By Aly | First published October 20, 2020 | Last modified November 29, 2022
Estrogens increase coagulation by activating estrogen receptors in the liver and thereby modulating the production of a variety of circulating coagulation factors. With sufficiently high exposure, this can result in an increase in the risk of blood clots as well as coagulation-associated cardiovascular complications like heart attack and stroke. However, the degrees of risk vary depending on the estrogen type, route, and dose. Non-bioidentical estrogens like ethinylestradiol have greater strength in the liver due to their relative resistance to metabolism and increase blood clot risk more readily than bioidentical estradiol, while oral administration of estradiol results in a first pass through the liver and has greater impact on blood clot risk than non-oral estradiol. Physiological estradiol levels with non-oral estradiol appear to have minimal to no risk of blood clots, whereas oral estradiol has significant risk and at high doses may have risk similar to that of the quantity of ethinylestradiol in modern birth control pills. Higher estradiol levels with non-oral estradiol seem to have significant risk of blood clots and cardiovascular problems as well, although the risks appear to be lower than with ethinylestradiol-containing birth control pills. Absolute risks of blood clots are low but accumulate with time and add up on a population scale. In addition, a variety of risk factors, such as age, physical inactivity, concomitant progestogen use, and often-unknown thrombophilic abnormalities, can substantially augment risk. Due to their higher risks of blood clots, oral estradiol as well as excessive doses of non-oral estradiol should ideally be avoided in transfeminine people. This is particularly applicable in those with risk factors for blood clots. In any case, therapeutic considerations for transfeminine people include not only safety but also effectiveness, other factors like cost and convenience, and the natures of the alternative therapeutic options.
Estrogens increase coagulation (blood clotting) and the risk of thrombosis, a cardiovascular event otherwise known as a blood clot. There are two major types of blood clots, which are categorized depending on whether they happen in a vein or in an artery: (1) venous thrombosis or venous thromboembolism (VTE); and (2) arterial thrombosis. VTE is a blood clot in a vein, a blood vessel that carries blood towards the heart. It comprises two different subtypes: (1) deep vein thrombosis (DVT), a clot in a vein of the leg or pelvic region; and (2) pulmonary embolism (PE), a clot that has broken free and blocked an artery in the lungs. Arterial thrombosis is a blood clot in an artery, a blood vessel that carries blood away from the heart. Arterial thrombosis can lead to myocardial infarction (MI; also known as heart attack) or cerebrovascular accident (CVA; also known as stroke). Blood clots are major health problems that can cause serious complications and even death. Estrogens, via increased coagulation with sufficiently high exposure, have the potential to heighten the risk of both venous and arterial thrombosis and hence to increase all of the aforementioned risks. The risk of blood clots with estrogens serves as a limiting factor in their use due to the potential health consequences.
Estrogens are selective agonists of the estrogen receptors (ERs). They are thought to increase coagulation and hence blood clot risk by activating ERs. However, the impact on coagulation and risk of blood clots with estrogens varies due to factors like estrogen type, route, and dose. In addition, other factors, like concomitant progestogen use and a variety of non-hormonal factors, are known to modify the risk. The purpose of this article is to review the risks of blood clots with estrogens, the mechanisms underlying increased coagulation and blood clot risk with estrogens, and the reasons for differences among estrogens in terms of risk. Exploring these topics can inform estrogen dosing considerations in transfeminine people and help to minimize risks and optimize safety. Moreover, higher levels of estrogens are therapeutically useful for suppressing testosterone production in transfeminine people but may increase blood clot risk, and risk–benefit analysis is warranted in this context.
A variety of estrogens have been used in medicine. These include bioidentical estrogens like estradiol (E2) as well as non-bioidentical estrogens like conjugated estrogens (CEEs; Premarin), ethinylestradiol (EE), and diethylstilbestrol (DES). Estradiol is the major natural estrogen in the human body. CEEs deliver primarily estradiol as the active estrogen, but also contain significant quantities of naturally occurring equine (horse) estrogens such as equilin (7-dehydroestrone) and 17β-dihydroequilin (7-dehydroestradiol). EE and DES are synthetic estrogens that were created by humans and do not occur naturally. DES was discontinued decades ago and is relatively little-known today, but has significant historical importance. Estradiol is used in both oral and non-oral forms (e.g., transdermal patches), while the non-bioidentical estrogens have typically been used orally. For context, the table below shows some approximate comparable doses of these estrogens in terms of general estrogenicity.
|Estrogen type/route||Very low dose a||Low dose a||Moderate dose b||High dose|
|Oral E2||1 mg/day||2 mg/day||4 mg/day||8 mg/day|
|Transdermal E2c||25 μg/day||50 μg/day||100 μg/day||200 μg/day|
|Oral CEEs||0.625 mg/day||1.25 mg/day||2.5 mg/day||5 mg/day|
|Oral EE||7.5 μg/day||15 μg/day||30 μg/day||60 μg/day|
|Oral DES||0.375 mg/day||0.75 mg/day||1.5 mg/day||3 mg/day|
|Comparable E2 level||~25 pg/mL||~50 pg/mL||~100 pg/mL||~200 pg/mL|
a Menopausal replacement dosages. b Similar to normal mean/integrated estrogenic exposure during the menstrual cycle in premenopausal women (Aly, 2018). c Specifically transdermal patches.
Estrogens were first associated with blood clots and associated cardiovascular complications in the 1960s and 1970s. Significant to substantial increases in these risks were found in clinical trials of high-dose DES (5 mg/day) for prostate cancer in men (VACURG, 1967; Byar, 1973; Turo et al., 2014), trials of moderate-dose CEEs (2.5–5 mg/day) for prevention of heart disease in men (Coronary Drug Project Research Group, 1970; Coronary Drug Project Research Group, 1973; Luria, 1989; Sudhir & Komesaroff, 1999; Dutra et al., 2019), and studies of early high-dose EE-containing birth control pills (50–150 μg/day) in premenopausal women (Gerstman et al., 1991; PCASRM, 2017; Table). The increase in cardiovascular events with DES in men with prostate cancer was sufficiently great that it actually cancelled out the benefits of its effects against prostate cancer in terms of overall mortality. The large increases in blood clots and cardiovascular problems seen in these studies resulted in alarm and concern about the safety of estrogens. Consequent to these events, estrogen doses were lowered. DES for prostate cancer was decreased to 1 to 3 mg/day and EE in birth control pills was decreased to 20 to 35 μg/day. Estrogens were also reduced to lower doses for other indications, such as menopausal hormone therapy. The dose reductions helped to lower the risks, although it did not eliminate them.
In the Women’s Health Initiative (WHI) randomized controlled trials (RCTs), low-dose oral CEEs alone (0.625 mg/day) were shown to slightly increase the risk of blood clots (Anderson et al., 2004; Curb et al., 2006; Prentice & Anderson, 2008; Prentice, 2014; Table). In addition, the increase was considerably augmented by concomitant use of a low dose (2.5 mg/day) of the progestogen medroxyprogesterone acetate (MPA) (Rossouw et al., 2002; Cushman et al., 2004; Prentice & Anderson, 2008; Prentice, 2014; Table). Increased risk of blood clots with low-dose oral CEEs plus low-dose MPA was also shown in another large RCT, the Heart and Estrogen/Progestin Replacement Study (HERS) (Hulley et al., 1998; Grady et al., 2000). Other progestogens besides MPA are also associated with augmentation of blood clot risk related to oral estrogens (Rovinski et al., 2018; Scarabin, 2018; Oliver-Williams et al., 2019; Vinogradova, Coupland, & Hippisley-Cox, 2019a; Table). Large observational studies have found low-dose oral estradiol (generally ≤2 mg/day) to be dose-dependently associated with increased risk of blood clots similarly to CEEs (Olié, Canonico, & Scarabin, 2010; Renoux, Dell’Aniello, & Suissa, 2010; Vinogradova, Coupland, & Hippisley-Cox, 2019a; Konkle & Sood, 2019; Table). However, the risk with oral estradiol or with oral esterified estrogens (a CEEs-like preparation with reduced equine estrogen content) appears to be lower than with oral CEEs (Smith et al., 2004; Smith et al., 2014; Vinogradova, Coupland, & Hippisley-Cox, 2019a; Table). On the other hand, in another large observational study, oral estradiol and oral CEEs both in combination with progestogens appeared to show similarly increased risk of blood clots (Roach et al., 2013). As with oral CEEs, progestogens appear to augment the blood clot risk with oral estradiol (Vinogradova, Coupland, & Hippisley-Cox, 2019a; Table).
In contrast to oral estrogens, transdermal estradiol at low to moderate doses (50–100 μg/day) has generally not been associated with increased coagulation nor with increased risk of blood clots or associated cardiovascular complications (Canonico et al., 2008; Hemelaar et al., 2008; Olié, Canonico, & Scarabin, 2010; Renoux, Dell’Aniello, & Suissa, 2010; Mohammed et al., 2015; Stuenkel et al., 2015; Bezwada, Shaikh, & Misra, 2017; Rovinski et al., 2018; Scarabin, 2018; Konkle & Sood, 2019; Oliver-Williams et al., 2019; Vinogradova, Coupland, & Hippisley-Cox, 2019a; Abou-Ismail, Sridhar, & Nayak, 2020; Table). Similarly, the Menopause, Estrogen and Venous Events (MEVE) study found that oral estradiol was associated with a large increase in risk of blood clots in women with previous history of blood clots whereas transdermal estradiol (mean dose 50 μg/day) was associated with no risk increase (Olié et al., 2011). However, there are some exceptions on findings of transdermal estradiol and cardiovascular risks, for instance one observational study finding an increased risk of stroke with higher-dose (>50 μg/day) transdermal estradiol patches in menopausal women (Renoux et al., 2010; Oliver-Williams et al., 2019) and studies finding only small differences or no difference in coagulation between oral estradiol and transdermal estradiol in transfeminine people (Lim et al., 2020; Scheres et al., 2021). Studies are mixed on whether the combination of transdermal estradiol at menopausal doses with progestogens is associated with greater blood clot risk, with some finding no change and others finding increased risk (Rovinski et al., 2018; Scarabin, 2018; Vinogradova, Coupland, & Hippisley-Cox, 2019a). It has been suggested that this may be related to the type of progestogen used (Scarabin, 2018).
There is little quality clinical data at this time on the risk of blood clots with higher doses of oral or transdermal estradiol than those used in menopausal hormone therapy. In any case, risk of blood clots has been assessed limitedly in transfeminine hormone therapy with regimens containing oral estradiol (e.g., 2–8 mg/day) generally in combination with other agents (antiandrogens and/or progestogens). In these studies, blood clot risk has been reported to be increased to a greater extent than with the low doses of oral estradiol used in menopausal hormone therapy (Wierckx et al., 2013; Weinand & Safer, 2015; Arnold et al., 2016; Getahun et al., 2018; Irwig, 2018; Connelly et al., 2019; Connors & Middeldorp, 2019; Goldstein et al., 2019; Iwamoto et al., 2019; Khan et al., 2019; Konkle & Sood, 2019; Quinton, 2019; Swee, Javaid, & Quinton, 2019; Abou-Ismail, Sridhar, & Nayak, 2020; Sam, 2020).
Whereas the WHI demonstrated causation for oral CEEs alone in terms of blood clot risk, no adequately powered RCTs have been conducted with oral or transdermal estradiol alone to establish causation in terms of blood clot risk at this time. Only very large and expensive trials would be able to show this due to the rarity of blood clots, and these studies have not been conducted to date. For similar reasons, RCTs demonstrating increased risk of blood clots with EE-containing birth control pills have also not been conducted at this time (Moores, Bilello, & Murin, 2004). In any case, causation has clearly been demonstrated with estrogens in other contexts, and this can be assumed as likely in the case of oral estradiol similarly. In addition, the Estrogen in Venous Thromboembolism Trial (EVTET), an RCT of low-dose (2 mg/day) oral estradiol plus the progestogen norethisterone acetate (NETA) versus placebo in postmenopausal women with history of previous blood clots, found that this hormone therapy regimen significantly increased coagulation and the incidence of blood clots (10.7% incidence with hormone therapy and 2.3% with placebo; P = 0.04) (Høibraaten et al., 2000; Høibraaten et al., 2001).
Estradiol levels appear to not be associated with blood clot risk in premenopausal women (Holmegard et al., 2014). The fact that transdermal estradiol patches at 100 μg/day in menopausal women haven’t been associated with a greater risk of blood clots is notable as this dose achieves estradiol levels of around 100 pg/mL on average, which are similar to the mean integrated levels of estradiol during the normal menstrual cycle in premenopausal women (Aly, 2018; Wiki). Rates of blood clots are also similar between men—who have relatively low estradiol levels—and women after controlling for atypical hormonal states like pregnancy and use of birth control pills in women (Moores, Bilello, & Murin, 2004; Rosendaal, 2005; Montagnana et al., 2010; Roach et al., 2013). Interestingly however, men have a consistently higher incidence of recurrent blood clots than women (Roach et al., 2013). These findings suggest that physiological levels of estradiol and progesterone in premenopausal women may not meaningfully increase coagulation or blood clot risk. However, the available data are mixed, with some studies suggesting that estradiol and/or progesterone levels within physiological ranges may indeed influence coagulation (Chaireti et al., 2013) and blood clot risk in premenopausal and/or perimenopausal women (Simon et al., 2006; Canonico et al., 2014; Scheres et al., 2019).
Modern combined birth control pills contain EE at moderately estrogenic doses (20–35 μg/day) and a physiological dose of a progestogen. They increase the risk of blood clots by several-fold (Konkle & Sood, 2019; Vinogradova, Coupland, & Hippisley-Cox, 2019b; Table). In addition, they are associated with about a 1.5- to 2-fold increase in risk of heart attack and stroke (Lidegaard, 2014; Konkle & Sood, 2019). However, overall mortality is not increased with birth control pills—at least in the relatively young women in whom they are used (Hannaford et al., 2010). Per studies of menopausal hormone therapy, it is likely that the progestogen in EE-containing birth control pills augments the risk of blood clots with EE. Early high-dose birth control pills (50–100 μg/day) had as much as twice the risk of blood clots of modern birth control pills (Gerstman et al., 1991; PCASRM, 2017; Table). In contrast to the different blood clot risks between oral and transdermal estradiol, non-oral birth control forms containing EE, for instance transdermal birth control patches and vaginal birth control rings, are associated with similar increases in blood clot risk as EE-containing birth control pills (Plu-Bureau et al., 2013; PCASRM, 2017; Konkle & Sood, 2019; Abou-Ismail, Sridhar, & Nayak, 2020). Hence, unlike with estradiol, route of administration does not appear to modify blood clot risk with EE based on available data.
High-dose estrogen therapy using oral synthetic estrogens like DES and EE in people with breast or prostate cancer has been found to strongly increase the risk of blood clots and associated cardiovascular complications (Phillips et al., 2014; Turo et al., 2014; Coelingh-Bennink et al., 2017). This has also been the case with estramustine phosphate (EMP; estradiol normustine phosphate), an estradiol ester that is used at massive doses in prostate cancer (e.g., 140–1,400 mg/day orally) and that results in pregnancy levels of estradiol (Kitamura, 2001 [Graph]; Ravery et al., 2011). In the 1980s however, it was found that high-dose non-oral estradiol did not have the same cardiovascular risks as high-dose estrogen therapy with oral synthetic estrogens or EMP (von Schoultz et al., 1989; Ockrim & Abel, 2009). This included studies with polyestradiol phosphate (PEP), a long-lasting injectable prodrug of estradiol, and with high-dose transdermal estradiol gel (von Schoultz et al., 1989; Aly, 2019). However, subsequent larger and higher-quality studies found that although the cardiovascular risks with PEP were much lower than with high-dose oral synthetic estrogen therapy, they were nonetheless still increased (Hedlung et al., 2008; Ockrim & Abel, 2009; Hedlund et al., 2011; Sam, 2020). This includes an approximate 2-fold increase in the risk of blood clots with estradiol levels in the range of roughly 300 to 500 pg/mL (Sam, 2020). Studies using high-dose transdermal estradiol patches have not found significantly increased cardiovascular complications as of present (Langley et al., 2013; Sam, 2020). However, these studies have been relatively underpowered, which limits their interpretation. In any case, increased coagulation has been observed with high-dose transdermal estradiol patches (achieving estradiol levels of 350 to 500 pg/mL) (Bland et al., 2005) similarly to PEP (Mikkola et al., 1999). More data on the risk of blood clots and cardiovascular issues with high-dose transdermal estradiol patches should come in the future with PATCH and STAMPEDE—two large-scale clinical studies in the United Kingdom that are evaluating this form of estradiol for prostate cancer (Gilbert et al., 2018; Singla, Ghandour, & Raj, 2019).
Injections of short-acting estradiol esters like estradiol valerate (EV) and estradiol cypionate (EC) are notable in that they are often used by transfeminine people and are generally used at doses that achieve high estradiol levels. As with high-dose transdermal estradiol patches, little to no quality data on the risk of blood clots exists for these preparations at present. Pyra and colleagues found that the risk of blood clots with injectable EV in transfeminine people was increased by around 2-fold, but the confidence intervals were very wide and statistical significance was not reached (Pyra et al., 2020). The doses used in the whole population for the study were not provided, but in the actual VTE cases, the doses of injectable EV were described and ranged from 4 to 20 mg once per week and 10 to 40 mg once every 2 weeks (Pyra et al., 2020). Studies have also assessed and found increased coagulation with high doses of EV by injection in the range of 10 to 40 mg once every 2 weeks in men with prostate cancer (Kohli et al., 2004; Kohli, 2005). Increased coagulation has additionally been observed with the combination of 5 mg EV and a progestogen once per month as a combined injectable contraceptive in premenopausal women (Meng et al., 1990; d’Arcangues et al., 2003). It is unclear whether the high peaks in estradiol levels associated with short-acting injectable forms of estradiol are harmful in terms of coagulation and blood clot risk (Hembree et al., 2017). However, the increased risk of polycythemia with short-acting injectable testosterone esters relative to other non-oral forms of testosterone (Ohlander, Varghese, & Pastuszak, 2018) is indirectly suggestive that this could be the case. Accordingly, a study found increased coagulation with a combined injectable contraceptive containing EV but not with one employing the more prolonged and stable EC at the same dose (d’Arcangues et al., 2003).
Selective estrogen receptor modulators (SERMs) such as tamoxifen (Nolvadex) and raloxifene (Evista) increase the risk of blood clots similarly to estrogens (Park & Jordan, 2002; Fabian & Kimler, 2005). The risk appears to be elevated a few-fold similarly to what might be expected with moderate doses of oral estradiol or CEEs (Deitcher & Gomes, 2004; Iqbal et al., 2012; Konkle & Sood, 2019).
Pregnancy is a time when estradiol and progesterone levels increase to extremely high concentrations (Graphs). Estradiol levels increase progressively throughout pregnancy to around 2,000 pg/mL on average at the end of the first trimester, to about 10,000 pg/mL on average at the end of the second trimester, and to around 20,000 pg/mL on average at the end of the third trimester (Kerlan et al., 1994 [Graph]; Schock et al., 2016). Coagulation is greatly increased during pregnancy, and the risk of blood clots is likewise strongly increased (Heit et al., 2000; Sultan et al., 2015; Heit, Spencer, & White, 2016; Table). Estradiol and progesterone levels are strongly correlated with the increases in coagulation during pregnancy (Bagot et al., 2019). The risk of blood clots with modern birth control pills is similar to that with pregnancy as a whole (Heit, Spencer, & White, 2016), while the increases in risk of blood clots with early high-dose EE-containing birth control pills and with high-dose oral synthetic estrogen therapy for breast and prostate cancer are comparable to the risk increase during late pregnancy. Estradiol levels also increase to very high concentrations during ovarian stimulation for in-vitro fertilization in premenopausal women, and this has been associated with increased coagulation and risk of blood clots as well (Westerlund et al., 2012; Rova, Passmark, & Lindqvist, 2012; Kasum et al., 2014).
Due to their greater risks of cardiovascular problems as well as other concerns, DES has been virtually abandoned while EE has been discontinued for almost all indications except birth control. EE continues to be used in birth control because it is resistant to metabolism in the uterus and controls menstrual bleeding better than oral estradiol does (Stanczyk, Archer, & Bhavnani, 2013). CEEs are also being increasingly superseded by estradiol in medicine, although significant use of CEEs for hormone therapy in cisgender women continues. Transdermal estradiol is gaining momentum over oral estradiol in menopausal hormone therapy as well. Major transgender hormone therapy guidelines (see also Aly, 2020) recommend against the use of EE and CEEs in transfeminine people due to their greater risks and the inability to accurately monitor blood estrogen levels with these preparations (Coleman et al., 2012; Deutsch, 2016; Hembree et al., 2017). Estradiol is the estrogen that is almost exclusively used in transfeminine people today. Besides estrogen type, it has been recommended that transdermal estradiol be used instead of oral estradiol in transfeminine people who are over 40 or 45 years of age or are otherwise at risk for blood clots (Deutsch, 2016; Iwamoto et al., 2019; Glintborg et al., 2021). Menopausal hormone therapy guidelines similarly recommend the use of transdermal estradiol over oral estrogens in cisgender women who are at higher risk for blood clots (e.g., Stuenkel et al., 2015).
As previously described, progestogens appear to augment the risk of blood clots with oral estrogens. Conversely, findings on the combination of non-oral estradiol and progestogens are mixed—with some studies finding increased risk and others finding no additional risk (Rovinski et al., 2018; Scarabin, 2018; Vinogradova et al., 2019). Progestogens by themselves do not usually increase coagulation (Kuhl, 1996; Schindler, 2003; Wiegratz & Kuhl, 2006; Sitruk-Ware & Nath, 2011; Sitruk-Ware & Nath, 2013; Skouby & Sidelmann, 2018) or blood clot risk (Blanco-Molina et al., 2012; Mantha et al., 2012; Tepper et al., 2016; Rott, 2019). However, depot MPA alone at birth control doses has uniquely been associated with a few-fold increase in blood clot risk (van Hylckama Vlieg, Helmerhorst, & Rosendaal, 2010; DeLoughery, 2011; Blanco-Molina et al., 2012; Gourdy et al., 2012; Mantha et al., 2012; Rott, 2019; Tepper et al., 2019). The reasons for this are unknown, but might relate to high peak MPA levels with depot injectables (Mantha et al., 2012) or the weak glucocorticoid activity of MPA (Kuhl & Stevenson, 2006; Sitruk-Ware & Nath, 2011). Besides physiological-dose MPA alone, high-dose progestogen therapy with MPA, megestrol acetate (MGA), and cyproterone acetate (CPA) has been associated with increased coagulation and blood clot risk (Schröder & Radlmaier, 2002; Schindler, 2003; Seaman et al., 2007; Garcia et al., 2013; Taylor & Pendleton, 2016). However, this was not the case with chlormadinone acetate (CMA) in a small study in women with prior history of blood clots (Conard et al., 2004). Risk of blood clots may also be increased for CPA in combination with estrogen in transfeminine people (Sam, 2020). In contrast to progestins, addition of oral progesterone to estrogen therapy is not associated with augmentation of blood clot risk (Scarabin, 2018; Oliver-Williams et al., 2019; Kaemmle et al., 2022). However, this may simply be due to the fact that oral progesterone produces low progesterone levels and has relatively weak progestogenic effects (Aly, 2018). Non-oral and fully potent progesterone has yet to be studied (Aly, 2018).
In a historically notable study conducted by the Center of Expertise on Gender Dysphoria (CEGD) at the Vrije Universiteit Medical Center (VUMC) in Amsterdam, the Netherlands in the 1980s, it was reported that the risk of blood clots with high-dose EE and CPA in transfeminine people was increased by 45-fold relative to the expected incidence in the general population (Asscheman, Gooren, & Eklund, 1989; Asscheman et al., 2014). Mortality also appeared to be elevated and other health risks were increased as well (Asscheman, Gooren, & Eklund, 1989; Gooren & T’Sjoen, 2018). A subsequent study in transfeminine people by the CEGD confirmed strongly increased coagulation with EE but much lower increases with oral or transdermal estradiol (Toorians et al., 2003). Upon the CEGD switching transfeminine people from high-dose EE to physiological doses of oral or transdermal estradiol (also usually in combination with CPA), the risks decreased considerably (van Kesteren et al., 1997; Asscheman et al., 2011; Asscheman et al., 2014). These findings were of major importance in the replacement of EE with estradiol in transfeminine hormone therapy, and have surely contributed significantly to apprehension about the use of high doses of estrogens in transfeminine people.
Taken together, estrogens of all kinds have been shown to dose-dependently increase or be associated with increased risk of blood clots. These findings suggest that, provided of course sufficient exposure occurs, increased coagulation and blood clot risk are common properties of estrogens. However, synthetic and non-bioidentical estrogens have greater risk of blood clots than estradiol, and oral estradiol shows greater risk than non-oral estradiol. In fact, physiological estradiol levels in women and low to moderate doses of transdermal estradiol may have no significant risk of blood clots at all. Nonetheless, non-oral estradiol with sufficiently high exposure can increase blood clot risk just the same as other forms of estrogen. Concomitant therapy with progestogens appears to augment the risk of blood clots with estrogens and high doses may particularly amplify the risk.
The table below provides relative risk increases for blood clots with different types, routes, and doses of estrogens, as well as with SERMs, pregnancy, and high-dose CPA. It shows the greater risks of blood clots with (1) oral estradiol relative to non-oral estradiol; (2) estradiol compared to non-bioidentical estrogens; and (3) lower estrogen levels/doses relative to higher estrogen levels/doses.
Table 2: Relative risks of blood clots with different hormonal exposures (see also Machin & Ragni, 2020):
|Estrogen||Blood clot risk||Source|
|Oral E2 ≤1 mg/day||1.2×||Vinogradova et al. (2019a) [Table]|
|Oral E2 >1 mg/daya||1.4×||Vinogradova et al. (2019a) [Table]|
|Oral E2 ≤1 or >1 mg/daya + Pb||1.4–1.8×||Vinogradova et al. (2019a) [Table]|
|Transdermal E2 ≤50 μg/day||0.9×||Vinogradova et al. (2019a) [Table]|
|Transdermal E2 >50 μg/daya||1.1×||Vinogradova et al. (2019a) [Table]|
|Oral CEEs ≤0.625 mg/day||1.4×||Vinogradova et al. (2019a) [Table]|
|Oral CEEs >0.625 mg/daya||1.7×||Vinogradova et al. (2019a) [Table]|
|Oral CEEs ≤ or >0.625 mg/daya + Pb||1.5–2.4×||Vinogradova et al. (2019a) [Table]|
|Modern EE + P birth controlc||4.2×||Heit, Spencer, & White (2016)|
|High-dose EE + P birth controlc||4–10×d||Tchaikovski, Tans, & Rosing (2006);|
PCASRM (2017) [Table]
|High-dose PEP injectionse||2.1×||Sam (2020)|
|High-dose oral DES, EE, or EMP||5.7–10×||Seaman et al. (2007); Ravery et al.|
(2011); Klil-Drori et al. (2015)
|SERMs (tamoxifen, raloxifene)||~1.5–3×||Deitcher & Gomes (2004); Iqbal et al.|
(2012); Konkle & Sood (2019)
|Pregnancy (overall)f||4.0×||Heit, Spencer, & White (2016)|
|Pregnancy (3rd trimester)||5.1–7.1×||Sultan et al. (2015) [Table]|
|High-dose CPA alone||3–5×||Seaman et al. (2007)|
a At typical menopausal replacement doses (i.e., not very high—probably no more than double the given dose). b MPA, norethisterone, norgestrel, or drospirenone. c Modern EE + P birth control contains 20 to 35 μg/day EE while high-dose EE + P birth control used in the 1960s and 1970s contained 50 to 150 μg/day EE. d Risk around twice as high as modern birth control pills. e Unpublished original research/analysis with borderline statistical significance (95% CI 0.99–4.22). f Excluding the postpartum period. With the postpartum period included, the risk of blood clots with pregnancy is 5–10× (McLintock, 2014).
Note that the values in the table are associations mostly from observational studies rather than from RCTs. Hence, in many cases, causation has not been definitively established. In addition, the values represent rough average values with often wide 95% confidence intervals. As a result, precision and accuracy of the estimates may in some cases be low. Also note that quantified blood clot risk will vary depending on the study and its definitions and methodology (including factors like sampling error, approach to control of confounding variables, and residual confounding influences).
The ERs are expressed in the liver and estrogens exert effects in this part of the body through these receptors (Eisenfeld & Aten, 1979; Eisenfeld & Aten, 1987; Sahlin & von Schoultz, 1999; Grossmann et al., 2019). Estrogens are thought to increase the risk of blood clots by activating liver ERs and thereby modulating the hepatic production of numerous different coagulation factors, both procoagulant and anticoagulant (Kuhl, 2005; Tchaikovski & Rosing, 2010; DeLoughery, 2011; Konkle & Sood, 2019). Most coagulation factors and their inhibitors are synthesized in the liver (Mammen, 1992; Amitrano et al., 2002; Peck-Radosavljevic, 2007). Following their synthesis, these coagulation factors are secreted by the liver into the bloodstream where they circulate and mediate their actions. Circulating levels of procoagulant factors like fibrinogen (factor I), prothrombin (factor II), factors VII, VIII, and X, anticoagulant factors like antithrombin, protein C, protein S, and tissue factor pathway inhibitor (TFPI), and fibrinolytic factors like plasminogen, tissue plasminogen activator (t-PA), and plasminogen activator inhibitor-1 (PAI-1), are all influenced by estrogens (Hemelaar et al., 2008; Doxufils, Morimont, & Bouvy, 2020). These estrogen-mediated changes in levels result in an overall procoagulatory effect, as assessed by markers of net coagulation activation like prothrombin fragment 1+2 (F1+2), D-dimer, and thrombin–antithrombin complex (TAT), as well as global coagulation assays like the endogenous thrombin potential-based activated protein C resistance test (The Oral Contraceptive and Hemostasis Study Group, 1999; Kohli, 2006; Hemelaar et al., 2008; Douxfils et al., 2020; Douxfils, Morimont, & Bouvy, 2020). The changes in levels of most coagulation factors caused by estrogens are relatively small and levels often remain within normal ranges. However, they combine and synergize to produce larger increases in global coagulation and clot risk (Douxfils et al., 2020; Douxfils, Morimont, & Bouvy, 2020; Reda et al., 2020).
Aside from coagulation factors, estrogens also modulate the synthesis of numerous other liver products (Kuhl, 1999; Kuhl, 2005; Table). Examples include sex hormone-binding globulin (SHBG), corticosteroid-binding globulin (CBG), various other circulating binding proteins, angiotensinogen, lipoproteins, and triglycerides, among others. In accordance with the mechanisms underlying increased coagulation and blood clot risk with estrogens, the differences in risk of blood clots with different types and routes of estrogens are mirrored in their influences on estrogen-sensitive liver products. Put another way, different estrogens have different relative potency in the liver when compared to their estrogenic potency elsewhere in the body. Synthetic and non-bioidentical estrogens have greater impact on liver synthesis than estradiol, while oral administration of estradiol has greater influence on liver synthesis than non-oral routes like transdermal administration or intramuscular injection, and this is likely to explain the observed differences in coagulation and blood clot risk with these different estrogens. The table below shows the liver potency of different estrogenic exposures as measured by influence specifically on SHBG levels, one of the most sensitive and well-characterized estrogen-modulated liver products.
Table 3: Relative increases in SHBG levels with different estrogenic exposures (see also Aly, 2020):
|E2 patch 50 μg/day||1.1×||Kuhl (2005)|
|E2 patch 100 μg/day||1.2×||Shifren et al. (2008)|
|Oral E2 1 mg/day||1.6×||Kuhl (1998)|
|Oral E2 2 mg/day||2.2×||Kuhl (1998)|
|Oral E2 4 mg/day||1.9–3.2×||Fåhraeus & Larsson-Cohn (1982); Gibney|
et al. (2005); Ropponen et al. (2005)
|Oral EV 6 mg/day (~4.5 mg/day E2)a||3.0×||Dittrich et al. (2005)|
|Oral CEEs 0.625 mg/day||1.8×||Kuhl (1998)|
|Oral CEEs 1.25 mg/day||2.2×||Kuhl (1998)|
|Oral EE 5 μg/day||2.0×||Kuhl (1999)|
|Oral EE 10 μg/day||3.0×||Kuhl (1998)|
|Oral EE 20 μg/day||3.4×||Kuhl (1998)|
|Oral EE 50 μg/day||4.0×||Kuhl (1997)|
|Modern EE + P birth controlb||~3.0–4.0×||Odlind et al. (2002)|
|High-dose EE + P birth controlb||~5–10×||Hammond (2017)|
|E2 patches 200 μg/day||~1.5×||Smith et al. (2019)|
|E2 patches 300 μg/day||~1.7×||Smith et al. (2019)|
|E2 patches 600 μg/day||~2.3×||Bland et al. (2005)|
|High-dose E2 injectionsc||1.7–3.2×||Stege et al. (1988); Mueller et al.|
(2011); Nelson et al. (2016)
|High-dose oral DES, EE, or EMPd||~5–10×||von Schoultz et al. (1989)|
a Due to differences in molecular weight, EV has about 75% of the amount of estradiol as regular estradiol. Hence, 6 mg/day EV is approximately equivalent to 4.5 mg/day E2. b Modern EE + P birth control contains 20 to 35 μg/day EE while high-dose EE + P birth control used in the 1960s and 1970s contained 50 to 150 μg/day EE. c In the form of 320 mg/month PEP (~700 pg/mL E2), 100 mg/month estradiol undecylate (EU) (~500–600 pg/mL E2), or 10 mg/10 days EV (~500–1,200 pg/mL peak E2; Graphs).
The increase in SHBG levels with estrogen therapy correlates with increases in coagulation and blood clot risk and can serve as a reliable surrogate indicator of these effects (Odlind et al., 2002; van Rooijen et al., 2004; van Vliet et al., 2005; Tchaikovski & Rosing, 2010; Raps et al., 2012; Stegeman et al., 2013; Hugon-Rodin et al., 2017; Eilertsen et al., 2019). The increases in SHBG levels and blood clot risk even appear quite similar to each other with modern birth control pills (both ~4-fold), high-dose oral synthetic estrogen therapy (both ~5–10-fold), and late pregnancy (both ~5–10-fold). When data on blood clot risk with a given estrogen route or dose are limited or unavailable—for instance with high-dose oral estradiol or high-dose estradiol ester injections—changes in SHBG levels can be used as a rough proxy or surrogate instead to estimate overall liver impact, magnitude of change in coagulation systems, and blood clot risk. It should be noted however that progestogens may augment the blood clot risk with estrogens without necessarily affecting SHBG levels or even while decreasing SHBG levels via concomitant androgenic activity (Kuhl, 2005; Vinogradova, Coupland, & Hippisley-Cox, 2019a).
Physiological levels of estradiol appear to have relatively minimal influence on liver synthesis (Eisenfeld & Aten, 1979; Lax, 1987; Kuhl, 2005). This is in accordance with the limited influence or non-influence of physiological estradiol levels in women on blood clot risk. It is thought that under normal physiological circumstances, estradiol is only supposed to considerably affect liver synthesis at very high levels—namely during pregnancy. The changes in synthesis of liver products during pregnancy presumably have important biological roles at this time (Eisenfeld & Aten, 1979). One of these is considered to be increased coagulation, as coagulation limits blood loss with childbirth and hence has survival benefits. Conversely, there is no obvious benefit to increased coagulation outside of pregnancy.
The oral route of administration is subject to a first pass through the liver via the hepatic portal vein which is not present with non-oral routes of administration (Pond & Tozer, 1984; Back & Rogers, 1987). As such, oral estradiol is subject to a hepatic first pass while this does not occur with non-oral forms of estradiol such as transdermal estradiol and injectable estradiol (Kuhl, 1998; Kuhl, 2005). This first pass results in disproportionate exposure of the liver to estradiol as well as disproportionate estrogenic impact on liver protein synthesis (Kuhl, 2005). Oral estradiol likewise has disproportionate estrogenic impact on the hepatic synthesis of coagulation factors (Kuhl, 1998; Kuhl, 2005). Due to the first pass, it is estimated that there is a 4- or 5-fold greater estrogenic impact of oral estradiol in the liver relative to non-oral estradiol (Kuhl, 2005). Due to the absence of the hepatic first pass with most non-oral routes, there is strong biological plausibility for the lower risk of blood clots that has been found with transdermal estradiol in comparison to oral estradiol in observational studies (Baber et al., 2016).
|Figure 1: Diagrammatic representation of increased coagulation via the liver first pass with oral estrogen therapy (Scarabin et al., 2020). Abbreviations: E = estrogen; trans = transdermal; AT = antithrombin; PS = protein S; TFPI = tissue factor protein inhibitor; II = prothrombin; VII = factor VII; PC = protein C; V = factor V; VTE = venous thromboembolism; CHD = coronary heart disease. Other terms: activated protein C resistance (APCR).|
Although oral estradiol has a much higher relative potential for blood clots due to the liver first pass, sufficiently high levels of estradiol will diffuse into the liver from the blood to act on this tissue regardless of route of administration. Hence, high levels of estradiol via non-oral routes (or produced by the body itself) can increase coagulation and blood clot risk similarly to the oral route. This is clearly evidenced by hyperestrogenic situations like pregnancy and ovarian stimulation for in-vitro fertilization, when estradiol levels increase to very high concentrations and substantially influence liver protein synthesis.
Non-bioidentical estrogens such as EE, DES, and CEEs have greater impact on liver protein synthesis and risk of blood clots than either oral estradiol or non-oral estradiol (Kuhl, 1998; Kuhl, 2005; Phillips et al., 2014; Turo et al., 2014; Table). This is because the liver strongly metabolizes and inactivates estradiol, whereas non-bioidentical estrogens have differences in their chemical structures relative to estradiol that result in them being much more resistant to liver metabolism (Kuhl, 1998; Kuhl, 2005; Connors & Middeldorp, 2019; Swee, Javaid, & Quinton, 2019).
EE can be considered as a case example. The oral bioavailability of EE is around 45%, while that of estradiol is only about 5% (Kuhl, 2005; Stanczyk, Archer, & Bhavnani, 2013). In addition, the blood half-life of EE is in the range of 5 to 30 hours, compared to less than 1 hour in the case of estradiol (White et al., 1998; Kuhl, 2005; Stanczyk, Archer, & Bhavnani, 2013). As a result of these and other differences, EE is approximately 120 times as potent as estradiol by weight in terms of general estrogenic effect (Kuhl, 2005; Table). Hence, EE is used clinically in μg doses whereas oral estradiol is used at over 100-fold higher mg doses. The pharmacokinetic differences between EE and estradiol reflect the strong resistance of EE to liver metabolism (Kuhl, 2005). EE, or 17α-ethynylestradiol, shows resistance to liver metabolism because of an ethynyl group at the C17α position which has been added to what is the otherwise unchanged structure of estradiol (Kuhl, 2005). This modification results in steric hindrance which blocks 17β-hydroxysteroid dehydrogenases (17β-HSDs) as well as conjugating enzymes like sulfotransferases and glucuronosyltransferases from metabolizing EE at the C17β hydroxyl group. 17β-HSDs normally convert estradiol into the weakly active estrone while the conjugating enzymes convert estradiol into inactive C17β estrogen sulfate and glucuronide conjugates like estrone sulfate (Kuhl, 2005). An “ethinylestrone” metabolite is in fact a structural impossibility due to the requirement of a double bond for a C17 ketone group—the needed C17α position is already occupied in EE by its ethynyl group. As such, the metabolism of estradiol into weakly active or inactive metabolites like estrone and estrone sulfate in the liver is protective against activation of hepatic ERs and procoagulation, and the lack of this with EE is responsible for its greater blood clot risk (Kuhl, 2005; Russell et al., 2017).
|Figure 2: Chemical structures of selected estrogens. The C17 position in the case of the steroidal estrogens (E2, E1, and EE) is at the top right of the steroid nucleus.|
Due to the marked resistance of EE to hepatic metabolism and inactivation, it persists for a long time in the liver—often cycling through it many times before finally being broken down. Moreover, EE shows several-fold disproportionate impact on liver protein synthesis at otherwise equivalent doses relative to oral estradiol (Kuhl, 2005; Table). Consequently, whereas EE has around 120-fold the general potency of oral estradiol, the liver potency of EE is around 350 to 1,500 times greater than that of oral estradiol (von Schoultz et al., 1989; Kuhl, 2005). A dose of EE of as little as 1 μg/day has been shown to impact liver metabolism (Speroff et al., 1996; Trémollieres, 2012). In addition, the fact that EE shows similar hepatic impact and risk of blood clots regardless of whether it is administered orally, transdermally, or vaginally indicates that unlike oral estradiol, the first pass through the liver with oral administration is not necessary for blood clot risk with EE (Plu-Bureau et al., 2013; PCASRM, 2017; Konkle & Sood, 2019). EE is so resistant to metabolism that it does not seem to matter how it is administered—the liver impact is substantial regardless of route. The greatly increased liver potency of EE results in its influence on coagulation and blood clot risk being much higher than that of estradiol at equivalent doses.
CEEs show a few-fold disproportionate estrogenic impact on liver protein synthesis relative to oral estradiol but less than that of EE (Kuhl, 2005; Table). This can be attributed to the equine (horse) estrogens in CEEs, which humans are presumably not adapted to and which show resistance to liver metabolism in humans. DES, on the other hand, shows even greater estrogenic influence on the liver than EE (Kuhl, 2005; Table). The more disproportionate impact on liver synthesis of DES relative to EE or CEEs may be attributable to the fact that it is a nonsteroidal estrogen and is far removed in structure from steroidal estrogens. This is relevant as steroidal estrogens are susceptible to varying extents to robust steroid-metabolizing enzymes in the liver (Kuhl, 2005). As with EE, 17β-HSDs have no affinity for DES and the hydroxyl groups of DES are not oxidized to form estrone-like ketone metabolites (Jensen et al., 2010). Consequent to their resistance to liver metabolism relative to estradiol, CEEs and nonsteroidal estrogens like DES have greater impacts on coagulation and blood clot risk than equivalent doses of estradiol similarly to EE although to varying extents.
When compared to transdermal estradiol rather than oral estradiol, the disproportionate influence of oral non-bioidentical estrogens on estrogen-modulated liver protein synthesis becomes extreme. With a little math, it quickly becomes apparent why high doses of these estrogens have influences on liver proteins and blood clot risks that are comparable to those during pregnancy. The table below shows some roughly calculated estimates for comparative liver strength of the different estrogens.
Table 4: Roughly calculated ratios of liver estrogenic potency to general/systemic estrogenic potency with estrogens based on a selection of liver products (e.g., SHBG, others) (Kuhl, 2005; Table):
|Estrogen||Comparative liver potency|
|Relative to oral E2||Relative to transdermal E2|
a Based on a study that found oral estradiol to have 4-fold greater effect on SHBG levels than transdermal estradiol when used at doses that produced similar estradiol levels (Nachtigall et al., 2000).
Changes in liver protein synthesis induced by estrogens don’t scale linearly with dose or relative liver potency. There is progressive saturation in terms of changes in levels of SHBG and other liver products with estrogen dose—that is, higher doses have relatively diminished effect compared to lower doses (Kuhl, 1990; Kuhl, 1999). As an example, oral EE shows the following dose-dependent increases in SHBG levels: 2.0-fold at 5 μg/day, 3.0-fold at 10 μg/day, 3.4-fold at 20 μg/day, and 4.0-fold at 50 μg/day (Kuhl, 1998; Kuhl, 1999). These findings can be attributed to saturation of the competitive binding and/or activation of liver ERs by high estrogen concentrations (Kuhl, 1990). An implication of this dose-dependent saturation is that although for instance oral EE has much stronger potency in the liver than oral estradiol, oral estradiol can more quickly “catch up” to oral EE and other non-bioidentical estrogens in terms of liver impact than might be initially anticipated. Accordingly, oral estradiol has shown the following dose-dependent increases in SHBG levels: 1.6-fold at 1 mg/day, 2.2-fold at 2 mg/day, and 1.9- to 3.2-fold at 4 mg/day (Fåhraeus & Larsson-Cohn, 1982; Kuhl, 1998; Gibney et al., 2005; Ropponen et al., 2005). Hence, although oral EE may have roughly 3- to 5-fold higher liver potency than oral estradiol, a dose of oral estradiol near-equivalent to that of oral EE in terms of general estrogenic effect can increase SHBG levels to an extent that is only somewhat lower in comparison.
SERMs like tamoxifen and raloxifene are essentially partial agonists of the ER. This is in contrast to estrogens—like estradiol, CEEs, EE, and DES—which act as full agonists of the ER. Similarly to nonsteroidal estrogens like DES, the clinically used SERMs are all nonsteroidal in structure and are strongly resistant to hepatic metabolism. In fact, certain SERMs, like tamoxifen and clomifene, are structurally related to and were derived from DES. SERMs show tissue differences in their ER-mediated effects, with estrogenic effects in some tissues (e.g., bone) and antiestrogenic effects in other tissues (e.g., breasts) (Lain, 2019; Table). Although there is variation between SERMs in terms of their effects in certain tissues (e.g., uterus), they are uniformly estrogenic in the liver. Consequently, SERMs show similar increases in blood clot risk as estrogens (Park & Jordan, 2002; Fabian & Kimler, 2005). As with non-bioidentical estrogens, the greater risk of blood clots with SERMs compared to oral estradiol can be attributed to their resistance to liver metabolism and hence to greater hepatic estrogenic potency. The SERMs that are used medically belong to diverse structural families (e.g., triphenylethylenes like tamoxifen and benzothiophenes like raloxifene). The only way in which SERMs of different structural classes are known to be related is in their shared interactions with the ERs.
|Figure 3: Chemical structures of selected SERMs. They are nonsteroidal in structure and include tamoxifen (a triphenylethylene) and raloxifene (a benzothiophene).|
Activation of the Estrogen Receptor is Specifically Responsible for Increased Coagulation with Estrogens and SERMs
Findings from preclinical and genetic research provide direct evidence for ER activation being responsible for the increased blood clot risk with estrogens. In an important animal study, EE was administered to mice and changes in procoagulant and anticoagulant biomarkers were measured (Cleuren et al., 2010). EE caused changes in levels of a variety of coagulation factors (Cleuren et al., 2010). The researchers also assessed estradiol and observed comparable changes (Cleuren et al., 2010). Co-administration of the selective ER full antagonist fulvestrant with EE neutralized all of the EE-induced coagulatory changes (Cleuren et al., 2010). Additionally, EE showed no effect on coagulation factors in ERα knockout mice (Cleuren et al., 2010). These findings are consistent with human and mouse genome-wide association studies which have found estrogen response elements (EREs)—DNA sequences that act as binding sites for genes regulated by the ER—embedded in a large number of genes involved in coagulatory pathways (Cleuren et al., 2010; Stanczyk, Mathews, & Cortessis, 2017).
The preceding findings are consistent with ER activation being responsible for increased coagulation and blood clot risk with estrogens and SERMs. This is in accordance with the fact that blood clot risk is a shared effect of selective ER agonists with highly diverse chemical structures, providing strong circumstantial support against a non-ER-mediated action of some sort being responsible (e.g., the weakly estrogenic metabolite estrone somehow mediating the blood clot risk with estradiol—Bagot et al., 2010). Increased coagulation and blood clot risk can thus be regarded as class effects of estrogens and SERMs—provided sufficiently high liver exposure. Due to differences in susceptibility to liver metabolism however, different ER agonists show differences in their relative impact on coagulation. Owing to estradiol’s lack of resistance to metabolism and its robust inactivation in the liver, the dosage requirements for increased coagulation and blood clot risk with estradiol—particularly in the case of non-oral estradiol—are greater than with non-bioidentical estrogens. Hence, estradiol, especially when administered via non-oral routes, is a safer form of estrogen therapy than other estrogens.
States of estrogen and/or progestogen exposure, such as exogenous hormone administration and pregnancy, are of course established risk factors for blood clots in women. In healthy young individuals without relevant risk factors for blood clots however, the incidence of blood clots is rare even in situations of considerably increased risk due to hormones (Rosendaal, 2005). The absolute incidence of VTE in non-pregnant women is only 1 to 5 of every 10,000 women each year (i.e., 0.01–0.05% per year) (PCASRM, 2017; Konkle & Sood, 2019). EE-containing birth control pills, which on average increase VTE risk by about 4-fold, are associated with an incidence of VTE of only 3 to 9 of every 10,000 women each year (i.e., 0.03–0.09% per year) (Konkle & Sood, 2019). Likewise, the absolute risk of blood clots during pregnancy, when estradiol and progesterone levels increase to extremely high concentrations and VTE risk is increased up to 7-fold (Sultan et al., 2015), is about 5 to 20 of every 10,000 women each year (i.e., 0.05–0.2% per year) (PCASRM, 2017; Konkle & Sood, 2019).
Table 5: Absolute incidences of VTE with different estrogenic exposures in premenopausal women (Gerstman et al., 1991; Konkle & Sood, 2019; Douxfils, Morimont, & Bouvy, 2020):
|Group/therapy||Incidence (women per year)|
|Non-pregnant women||1 to 5 in 10,000 (0.01–0.05%)a|
|Modern birth control pills (<50 μg/day EE)||3 to 12 in 10,000 (0.03–0.09%)|
|High-dose birth control pills (>50 μg/day EE)||~10 in 10,000 (0.1%)|
|Pregnancy||5 to 20 in 10,000 (0.05–0.2%)|
|Postpartum period||40 to 65 in 10,000 (0.4–0.65%)|
a 1–2/10,000 per year at <19 years of age, 2–3/10,000 per year at 20–29 years of age, 3–4/10,000 per year at 30–39 years of age, 5–7/10,000 per year at 40–49 years of age; roughly 3–4/10,000 per year for age 15–49 years overall (Rabe et al., 2011).
In any case, the risks of VTE and cardiovascular events with high estrogen exposure accumulate over time and add up on a population scale. It is estimated that 22,000 instances of VTE occur due to birth control pills in Europe each year (Morimont, Dogné, & Douxfils, 2020) and that 300 to 400 healthy young women die due to blood clots caused by birth control pills in the United States every year (Keenan, Kerr, & Duane, 2019). Notably, non-EE-containing birth control pills—which instead of EE contain estradiol or estetrol—appear to have considerably reduced procoagulatory effects and/or risk of blood clots in comparison, and if they become more established, will likely eliminate a substantial number of these cases (Stanczyk, Archer, & Bhavnani, 2013; Dinger, Minh, & Heinemann, 2016; Grandi, Facchinetti, & Bitzer, 2017; Fruzzetti & Cagnacci, 2018; Grandi et al., 2019; Grandi et al., 2020; Douxfils, Morimont, & Bouvy, 2020; Reda et al., 2020; Morimont et al., 2021; Grandi, Facchinetti, Bitzer, 2022).
In addition to time and population considerations, there are, besides estrogen and progestogen exposure, a variety of other known risk factors for blood clots, and these risk factors can substantially augment blood clot risk (Heit et al., 2000; Rosendaal, 2005). Age is among the strongest of the known risk factors (Rosendaal, 2005; Montagnana et al., 2010). Moreover, age is uniquely notable as a risk factor in that it is one that eventually becomes relevant to everyone. The risk of blood clots increases on the order of 100-fold going from ≤15 years of age (incidence <0.005–0.01% per year) to ≥80 years of age (incidence ~0.5–1.0% per year) (Rosendaal, 2005; Montagnana et al., 2010; Rabe et al., 2011). The figure below provides a graphical representation of the influence of age on risk of blood clots.
|Figure 4: Risk of first-incidence VTE (per 100,000 per year) by age group (in years) in men (black bars) and women (gray bars) (Oger, 2000; Rosendaal, 2005; Rosendaal, 2016).|
Other established risk factors for blood clots and associated cardiovascular problems include physical inactivity (due to, e.g., bed rest, long-distance travel, etc.), obesity, smoking, thrombophilic abnormalities, cancer, surgery, and HIV, among many others (Baron et al., 1998; Heit et al., 2000; Rosendaal, 2005; Lijfering, Rosendaal, & Cannegieter, 2010; Timp et al., 2013). In addition to age, physical inactivity is one of the most important risk factors for blood clots and mediates the risk increases for many of the others (Rosendaal, 2005). Smoking on its own is not consistently associated with increased risk of VTE (Lijfering, Rosendaal, & Cannegieter, 2010), but in combination with EE-containing birth control pills has been associated with a synergistic increase in VTE risk (Pomp, Rosendaal, & Doggen, 2008) as well as large increases in risk of heart attack—for instance 20-fold higher risk in heavy smokers (Kuhl, 1999). The table below shows the influence of a selection of known risk factors for VTE:
Table 6: Non-exogenous-hormone risk factors for VTE and relative VTE risk increases (Baron et al., 1998; Heit et al., 2000; Rosendaal, 2005; Lijfering, Rosendaal, & Cannegieter, 2010; Timp et al., 2013):
|Risk factor||Relative risk|
|Surgery, trauma, immobilization||5–50×|
|Bed rest at home||9×|
a Varies by type and stage of cancer (Baron et al., 1998; Timp et al., 2013). For breast and prostate cancer, one study found a 1.8-fold greater risk for breast cancer and 4.2-fold greater risk for prostate cancer relative to the general population (Baron et al., 1998). b Smoking on its own is not consistently associated with VTE (Lijfering, Rosendaal, & Cannegieter, 2010; Rabe et al., 2011).
Thrombophilias, heritable and acquired, exist in significant percentages of the population and can lead to large increases in blood clot risk (Lijfering, Rosendaal, & Cannegieter, 2010). Moreover, they are often if not usually unknown (Morimont, Dogné, & Douxfils, 2020). This is due to the fact that screening for heritable thrombophilias is mainly based on family history, which has low sensitivity and poor predictive value for identifying people with these abnormalities (Morimont, Dogné, & Douxfils, 2020). Hence, many people are at increased risk of blood clots without realizing it. The table below shows the prevalences of a variety of thrombophilic abnormalities and their impacts on blood clot risk.
|General population||People with VTE||First VTE||Recurrent VTE|
|Protein C deficiency||0.2–0.4%||3%||15×||2.5×|
|Protein S deficiency||0.03–0.1%||2%||10×||2.5×|
|Factor V Leiden (het.)||5%||20%||7×||1.5×|
|Factor V Leiden (homo.)||0.02%||1.5%||80×||–|
|Prothrombin G20210A (het.)||2%||6%||3–4×||1.5×|
|Prothrombin G20210A (homo.)||0.02%||<1%||30×||–|
|Non-O blood group||55–57%||75%||2×||2×|
Blood clots are considered to be a multicausal disease (Rosendaal, 2005). The risk of blood clots and associated cardiovascular complications with hormonal exposure is highest when multiple risk factors combine in a given individual. Under what are among the most extreme of circumstances in terms of risk—elderly people with cancer who are on high-dose oral synthetic estrogen therapy (e.g., DES)—blood clot incidence can be as high as 15 to 28% and overall incidence of cardiovascular complications as great as 35% (Phillips et al., 2014; Sciarria et al., 2014; Turo et al., 2014). These adverse effects contribute to substantial morbidity and incidence of death in these populations. Most people are however at nowhere near as great of risk. Risk factors like age are why pregnant women can have massive levels of estradiol and progesterone with relatively little issue whereas elderly cancer patients on high-dose oral synthetic estrogen therapy have a considerable risk of death.
In the VUMC studies that found 20- to 45-fold increased incidence of blood clots with high-dose EE and CPA over 5 to 10 years in transfeminine people, the absolute incidence of blood clots was approximately 6.3% (142/10,000 people per year) in the 1989 report and 5.5% (58/10,000 people per year) in the 1997 follow up (Asscheman, Gooren, & Eklund, 1989; van Kesteren et al., 1997; Asscheman et al., 2014; Goldstein et al., 2019; Min & Hopkins, 2021). In keeping with the known influence of age on blood clot risk, the absolute incidence was 2.1% in those under 40 years of age and 12% in those over 40 years of age in the 1989 study (Asscheman, Gooren, & Eklund, 1989; Asscheman et al., 2014). In about 70% of cases, there were—aside from age—no known risk factors for blood clots (Asscheman, Gooren, & Eklund, 1989; Asscheman et al., 2014). Following subsequent replacement of EE with low-to-moderate-dose transdermal estradiol in those over 40 years of age, the incidence of blood clots decreased substantially (with only one event occurring in the transdermal estradiol group) (van Kesteren et al., 1997; Asscheman et al., 2014; Min & Hopkins, 2021). A later study in 2013 by the Ghent University Hospital in Belgium observed a blood clot incidence of 5.1% in transfeminine people using mostly oral or transdermal estradiol with or without CPA over an average treatment period of 7.7 years (range 3 months to 35 years) (Wierckx et al., 2013; Min & Hopkins, 2021). Those who had blood clots often had other risk factors such as older age, smoking, immoblization due to surgery, or hypercoagulability (Wierckx et al., 2013; Min & Hopkins, 2021). In addition to cumulative exposure time, these studies further highlight the converging impact of multiple risk factors—with estrogen type, route, and dose, progestogen exposure, and age included among them—on the risk of blood clots.
Due to their greater risk of blood clots and cardiovascular problems, non-bioidentical estrogens like EE and CEEs are mostly no longer used in transfeminine people. Instead, estradiol, both in oral and non-oral forms, is used. Transgender clinical guidelines generally recommend keeping estradiol levels within normal physiological ranges for non-pregnant females of around 100 to 200 pg/mL regardless of whether the route of administration of estradiol is oral or non-oral (Aly, 2018). Higher estradiol levels are not currently known to have greater therapeutic benefit in terms of feminization or breast development (Nolan & Cheung, 2020). However, higher levels, in the range of 200 to 500 pg/mL, can provide additional therapeutic effect in the area of testosterone suppression—which can be indirectly beneficial to feminization if otherwise inadequate (Aly, 2018). Despite their recommendations for keeping estradiol levels in physiological ranges, transgender clinical guidelines notably recommend doses of estradiol ester injections that reach and even greatly exceed estradiol levels of 200 pg/mL (Aly, 2021).
Based on the available research (e.g., the risk of blood clots with lower doses, comparative SHBG increases), it would not be surprising if high-dose oral estradiol (e.g., 8 mg/day) had similar risk of blood clots as the relatively lower amounts of EE in birth control pills. The risk is likely to be particularly great in combination with progestogens (e.g., CPA). Due to its greater and unnecessary risk of blood clots relative to non-oral estradiol, oral estradiol should ideally be avoided in transfeminine people—particularly in those with risk factors for blood clots such as older age (e.g., >40 years) or concomitant progestogen use. However, the convenience of oral estradiol and its relative inexpensiveness (compared to e.g. transdermal forms) are significant advantages that will also be considered by transfeminine people and their clinicians. In contrast to oral estradiol, non-oral estradiol—with estradiol levels kept in physiological ranges of for instance 100 to 200 pg/mL—appears to have minimal to no risk of blood clots. Hence, non-oral estradiol at these levels can be used in transfeminine people with little concern.
In terms of higher estradiol levels delivered non-orally, the estimated 2-fold increase in risk of blood clots with estradiol levels of approximately 300 to 500 pg/mL (Sam, 2020) is notably lower than the average 4-fold increase in risk with widely used EE-containing birth control pills. Based on the usefulness of these levels for suppressing testosterone production and the widespread usage of EE-based birth control in cisgender women throughout the world, the degree of blood clot risk with high-dose non-oral estradiol—in reasonable amounts—could be considered therapeutically acceptable in transfeminine people (Haupt et al., 2020). This may be particularly true when high-dose non-oral estradiol monotherapy is compared to alternative regimens like estradiol plus antiandrogens including spironolactone, CPA, or bicalutamide—which all have their own unique risks and drawbacks. In fact, at least in the specific case of estradiol plus CPA, the risks of this particular regimen are probably worse in comparison (e.g., Sam, 2020; Aly, 2020; Aly, 2020). In any case, as with oral estradiol, high estradiol levels with non-oral estradiol should ideally be avoided due to the additional risk they pose, and this is especially true in those with relevant risk factors for blood clots (e.g., older age). In addition, very high doses of non-oral estradiol resulting in estradiol levels above those required for testosterone suppression are difficult to justify as they pose further unnecessary risk and offer no clear additional therapeutic benefit.
The best way to prevent blood clots from happening is to avoid risk altogether. Avoiding use of oral estradiol, excessively high doses of non-oral estradiol, and progestogens when feasible and opting for safer therapeutic choices is recommended in this regard. In addition, avoiding use of such therapies in those with risk factors like older age (>40 years), known thrombophilic abnormalities, and sedentary lifestyle is advocated. Proactive behaviors like physical activity (e.g., walking, exercise), quitting smoking, and weight loss may help to reduce the risk of blood clots (Hibbs, 2008).
Certain anticoagulant and antiplatelet medications are used to help prevent blood clots in high-risk individuals. Examples include low-dose aspirin (Mekaj, Daci, & Mekaj, 2015; Matharu et al., 2020), direct factor Xa inhibitors like rivaroxaban (Xarelto) (Blondon, 2020), and direct thrombin inhibitors like dabigatran (Pradaxa), among others. Aspirin has been found to be effective in the prevention of blood clots (Mekaj, Daci, & Mekaj, 2015; Matharu et al., 2020) and has been recommended for use specifically in transfeminine people on hormone therapy (Feldman & Goldberg, 2006; Deutsch, 2016). However, evidence is limited and conflicting for prevention of blood clots related to hormone therapy (Grady et al., 2000; Cushman et al., 2004) and use of aspirin in transfeminine people for such purposes has been recommended against by others (Shatzel, Connelly, & DeLoughery, 2017). Rivaroxaban has been associated with more than completely offset risk of blood clots with oral menopausal hormonal therapy (Blondon, 2020). In any case, no anticoagulants are currently approved or well-supported for preventing risk of blood clots with hormone therapy. Accordingly, clinical guidelines state that there is insufficient evidence to guide decision-making in this area at this time (e.g., McLintock, 2014). It should also be cautioned that anticoagulants have side effects and risks of their own and should be used carefully.
Rutin, a naturally occurring flavonoid found in various plants and foods and available as a herbal supplement, has been suggested by some in the transfeminine community as a preventative against blood clots based on limited preclinical research (Jasuja et al., 2012; Choi et al., 2015). However, there is no clinical evidence to support its use or effectiveness at this time (e.g., Martinez-Zapata et al., 2016; Morling et al., 2018). Dose-finding studies to determine appropriate doses for efficacy also have not been performed. Flavonoids like rutin are notably known to have unfavorable dispositions in the body (e.g., very low bioavailability, high metabolism, short half-lives) and this has limited their usefulness by rendering them poorly active and therapeutically ineffective (Ma et al., 2014; Higdon et al., 2016; Cassidy & Minihane, 2017; Zhao, Yang, & Xie, 2019; Zhang et al., 2021). Lastly, the tolerability and safety of rutin have not been evaluated. For these reasons, use of rutin to lower the risk of blood clots in transfeminine people cannot be recommended at this time.
Temporary discontinuation of estrogen therapy before surgery has traditionally been thought to help reduce the risk of blood clots during recovery based on theory and has been advised as well as mandated for transfeminine people undergoing surgical procedures (e.g., Asscheman et al., 2014). However, evidence is limited and inconclusive on this strategy at present and more research is needed to determine whether it is actually beneficial or not (Boskey, Taghinia, & Ganor, 2019; Nolan & Cheung, 2020; Haveles et al., 2021; Hontscharuk et al., 2021; Kozato et al., 2021; Nolan et al., 2021; Zucker, Reisman, & Safer, 2021). Recent studies have not found reduction in risk of blood clots with discontinuation of hormone therapy before surgery in transfeminine people but these studies have been underpowered and larger studies are needed (Blasdel et al., 2021). Temporarily stopping hormone therapy can be distressing for many transfeminine people and this should be weighed accordingly. A potential alternative to discontinuation of hormone therapy is temporary use of transdermal estradiol at physiological doses which has no known blood clot risk and is more likely to be safe.
In February 2021, a report on long-term cardiovascular outcomes for the Prostate Adenocarcinoma: TransCutaneous Hormones (PATCH) trial was published (Langley et al., 2021) [PDF; Supplementary appendix]. The PATCH trial is a large ongoing phase 2/3 randomized controlled trial of high-dose transdermal estradiol patches versus GnRH agonists for the treatment of prostate cancer in men (Langley et al., 2021). The estradiol patch dosage employed is specifically three to four 100 μg/day FemSeven or Progynova TS patches (Langley et al., 2021). In the February 2021 report of the study, 1,694 men were enrolled and randomized, with 790 included in the analysis for the GnRH agonist group and 904 included in the analysis for the estradiol patch group (Langley et al., 2021).
In those given estradiol, the median estradiol level was around 215 pg/mL (5%–95% range ~100–550 pg/mL) (Langley et al., 2021). About 93% of the men in this group achieved suppression of testosterone levels into the castrate range (<50 ng/dL), which was notably equal to the rate of suppression in the GnRH agonist group (~93%) (Langley et al., 2021). However, actual testosterone levels—as opposed to rates of testosterone suppression—were not provided in this report and hence comparison between groups is unavailable for this metric (Langley et al., 2021). After about 4 years median follow up, there were no significant differences on a variety of cardiovascular outcomes between the estradiol group and the GnRH agonist group (Langley et al., 2021). Among these outcomes included VTE, thromboembolic stroke, and other arterial embolic events (Langley et al., 2021). These results are in contrast to previous large clinical trials of PEP in prostate cancer, which found increased cardiovascular morbidity and risk of VTE but notably involved higher estradiol levels than employed in the PATCH trial (Ockrim & Abel, 2009; Sam, 2020). Based on their promising safety findings, the PATCH researchers stated that transdermal estrogen should be reconsidered for the treatment of prostate cancer (Langley et al., 2021).
These findings are reassuring and suggest that limitedly high levels of estradiol (e.g., 200–300 pg/mL perhaps) may likewise be acceptably safe in terms of blood clot and cardiovascular risk in transfeminine people. It should be noted however that the sample size of the trial, while large relative to previous clinical studies in this area, was underpowered for assessing risk of blood clots—which are relatively rare events that require very large samples to thoroughly quantify. Studies precisely assessing blood clot risk in peri- and postmenopausal women have included tens of thousands of individuals for instance. As such, while substantial increases in risk are not likely based on this trial, smaller increases in risk still cannot be ruled out at this time. It should additionally be noted that the robust testosterone suppression at the used doses in this study might not generalize to transfeminine people as a whole, as the men were mostly elderly and testosterone levels are known to decrease with age.
- Totaro, M., Palazzi, S., Castellini, C., Parisi, A., D’Amato, F., Tienforti, D., Baroni, M. G., Francavilla, S., & Barbonetti, A. (2021). Risk of Venous Thromboembolism in Transgender People Undergoing Hormone Feminizing Therapy: A Prevalence Meta-Analysis and Meta-Regression Study. Frontiers in Endocrinology, 12, 741866. [DOI:10.3389/fendo.2021.741866]
This study is the largest of its kind that has been conducted to date. The meta-analysis included 18 studies totaling 11,542 transfeminine people on hormone therapy. The pooled prevalence of VTE was 2% with a 95% confidence interval of 1 to 3%. However, there was large variability between studies. In the meta-regression analysis, older age and longer length of estrogen therapy were significantly positively associated with VTE prevalence. When analysis was restricted to those greater than or equal to 37.5 years of age, the prevalence of VTE was 3% (95% CI: 0–5%). Conversely, in those less than 37.5 years of age, the prevalence of VTE was 0% (95% CI: 0–2%). VTE prevalence was 1% (95% CI: 0–3%) with greater than or equal to 4.4 years of estrogen therapy and was 0% (95% CI: 0–3%) with less than 4.4 years of estrogen therapy. With regard to the 0% estimates, it is not the case that there is truly no risk of VTE in these instances but rather it can be assumed that the risks are sufficiently low that the meta-analysis was not powered well enough to detect and quantify them.
A limitation of the meta-analysis was that subgroup analyses based on estrogen type (i.e., estradiol vs. CEEs vs. EE) and route (e.g., oral estrogens or oral estradiol vs. transdermal estradiol) were said to not be possible due to insufficient data and hence were not performed. However, another recent meta-analysis published in July 2021, which analyzed much of the same literature as Totaro et al. (2021), did perform subgroup analyses by estrogen type and route. This publication is as follows:
- Kotamarti, V. S., Greige, N., Heiman, A. J., Patel, A., & Ricci, J. A. (2021). Risk for Venous Thromboembolism in Transgender Patients Undergoing Cross-Sex Hormone Treatment: A Systematic Review. The Journal of Sexual Medicine, 18(7), 1280–1291. [DOI:10.1016/j.jsxm.2021.04.006]
And this is what they reported in terms of subgroup analyses for estrogen type and route:
Because varying VTE rates have been reported with different estrogen regimens, analyses of VTE incidence were performed comparing oral or transdermal delivery, or the specific estrogen formulation. As many studies reported populations using mixed estrogen formulations or did not report the type of estrogen regimen, further statistical analysis could not be performed.
Route of estrogen administration appeared to play a role in the AMAB population. [Oral] estrogens (7 studies; 34.0 VTE per 10,000 person-years) vs transdermal estrogens (3 studies, 11.2 VTE per 10,000 person-years). Additionally, estrogen formulation also appeared to have a difference VTE incidence. Ethinyl estradiol was also associated with increased VTE incidence (3 studies, 293.1 VTE per 10,000 person-years) followed by conjugated equine estrogens (1 study, 49.0 VTE per 10,000 person-years) and estradiol valerate (4 studies, 31.5 VTE per 10,000 person-years).
It is unclear how accurate these precise numbers are due to the quality limitations of the underlying data. Moreover, antiandrogens (e.g., CPA) were not controlled for and as discussed by this article are likely to additionally influence VTE risk. In any case, the reported numbers are interesting and are in accordance with different estrogen types and routes varying in terms of VTE risk.