Hormones and thrombosis: risk across the reproductive years and beyond.

PCOS is a disorder characterized by androgen excess. The global disease prevalence ranges from 5% to 20% among premenopausal women and, as such, PCOS is one of the most common endocrine system and reproductive disorders in women. 21 Typical signs and symptoms include irregular to no menses, hirsutism, alopecia, acne, and ovulatory dysfunction. Cysts are present in the ovaries and detectable by ultrasound, and while central to the pathology of the disease, and the ovarian follicle theca cells produce androgens, ovarian cysts are not the cause of PCOS. In fact, the cause of PCOS is not well understood, and is classified as a complex, multigenic disorder, 22 with a myriad of effects on reproductive, metabolic, and cardiovascular systems. The management of PCOS is with low-dose estrogen combination oral contraceptive to regulate the cycle and reduce acne and hyperandrogenism. 15 , 21 Women with PCOS are at increased risk for VTE, with a 1.5–2.0 fold greater than in unaffected women ( Table 1 ). 22 The VTE risk in women with PCOS appears to be independent of clinical obesity and use of combination oral contraceptives. 23 , 24 In fact, a population-based cohort study reported that use of combination oral contraceptives increased VTE risk in women with PCOS nearly 4-fold, 23.7/10,000 person-years vs 6.3/10,000 person-years, compared to women with untreated PCOS. 25 For most patients, however, the benefits of combination hormonal therapy, though, outweigh the small increase in absolute VTE risk, but this remains an important risk-benefit consideration in their management. The VTE risk in women with PCOS is attributable to the endogenous production of estradiol, sex-hormone-binding globulin, and free androgen index, each of which is individually associated with increased VTE risk. 26 They also have a hypofibrinolytic state associated with greater levels of plasminogen activator inhibitor 1 antigen and thrombin activatable fibrinolysis inhibitor, 27 , 28 which each inhibit fibrinolysis and promote thrombosis. 29 In women who develop acute VTE, standard anticoagulation guidelines apply, including UFH or LWMH for acute thrombosis, followed by 3–6 months of LWMH or DOACs ( Tables 3 and 4 ). 17 – 19 In those who develop VTE while on hormonal therapy, continued use may require ongoing anticoagulation, although there are few studies to guide management.

Breast cancer is the most common cancer and the second leading cause of cancer death in women in the United States. 73 VTE has been shown to increase with age in women with breast cancer. 74 Hormonal therapy for premenopausal breast cancer, including clomiphene or aromatase inhibitors, is associated with low risk for VTE. In estrogen receptor-positive premenopausal breast cancer, the frequency of VTE after 5 years of tamoxifen therapy was 2.2% (95% CI, 1.4, 3.3), 75 and even after extending treatment to 10 years, the risk remained low, RR 1.87 [1.13, 3.07] ( Table 1 ). 76 By contrast, postmenopausal breast cancer treated with antihormonal therapy, including tamoxifen or aromatase inhibitors, has been associated with greater rates of VTE. In the multicenter, prospective, longitudinal COMPASS-CAT study (Prospective Comparison of Methods for thromboembolic risk assessment with clinical perceptions and Awareness in real life patients-Cancer Associated Thrombosis), 77 those receiving antihormone therapy for breast cancer had over a 3-fold increased risk for VTE compared with women not receiving antihormone therapy, OR 2.64 [0.82, 2.23]. In multivariate analysis, VTE was the strongest predictor of antihormonal therapy was associated with the highest risk of any other risk factor for VTE, OR 6.4 [3.16, 12.96]. 77 Selective estrogen receptor modulators (SERMs) are nonsteroidal agents with tissue-specific effects used in postmenopausal women. 73 , 78 They show beneficial effects on breast, bone, the cardiovascular system, endometrium, vagina, and vasomotor symptoms. 78 SERMs have varying agonist and antagonist activities in different tissues, relieving symptoms of menopause, eg, hot flashes and dyspareunia, and preventing bone loss and osteoporosis, while maintaining antagonistic or neutral effects on breast and endometrial tissue. 78 There has been, however, no clear evidence of cardio-protection or neuroprotection with SERMs. As a class, SERMs appear to increase VTE risk 2-fold, similar to risk with estrogens, with no difference between SERMs ( Table 1 ). 78 – 80 While first line therapy for acute cancer-associated thrombosis is LWMH followed by LMWH prophylaxis, 81 2 recent randomized trials in cancer patients with acute VTE, the Hokusai VTE Cancer study 82 and the SELECT-D trial, 83 have established that DOACs have the same or better efficacy than LWMH. Yet, DOACS may have a higher risk of bleeding, and, in patients with thrombocytopenia <50,000/L complicating chemotherapy, LMWH may be favored over DOACs ( Tables 3 and 4 ).

Since many women choose to delay childbearing, the rates of infertility and the utilization of assisted reproductive technology (ART), which includes both ovulation-induction therapy and IVF, have increased. The CDC reported 280,000 cycles of IVF occurring in US clinics, which resulted in 68,000 live births in 2017. 45 Ovulation-induction therapy to promote multiple follicle development within the ovary is a central part of IVF cycles. However, for the vast majority of women, IVF is safe and serious complications are rare. However, ovulation induction may be associated with supraphysiologic levels of estrogens, 46 increased levels of coagulation factors and enhanced thrombosis risk which may be observed in association with ovarian hyperstimulation syndrome (OHSS). 47 , 48 The use of gonadotropin releasing hormones and gonadotropin releasing hormone agonists, instead of exogenous human chorionic gonadotropin, to trigger ovulation has greatly reduced VTE and OHSS. Data from the large RIETE registry, a multi-center prospective study of patients with VTE, the risk for VTE in women undergoing ART was 4.13 [1.4, 12.4] ( Table 1 ), 49 and among women undergoing unsuccessful IVF, VTE risk was more common, OR 5.0 [1.2, 20.7]. 49 These rates are significantly higher than those during contraception or pregnancy. In a systematic review of the literature, the incidence of VTE after IVF was found to be 0.8/1000 in women without OHSS, compared to 25/1000 with OHSS, while the control pregnant population had an incidence of 0.17–2.5/1000. 50 Based on these findings, the authors concluded the antepartum risk of VTE after IVF is approximately doubled, with the greatest risk occurring in the first trimester. 51 While most of the risk associated with IVF is in conjunction with cycles of successful ovulation, an evaluation of RIETE registry data identified an increased risk of pulmonary embolism (PE) in women undergoing unsuccessful IVF as well (OR: 5.0, 95% CI 1.2–20.7). 49 The current guidelines to prevent VTE in women undergoing ART are based on the guidelines to prevent VTE in pregnancy. 20 , 48 , 52 , 53 LMWH is the anticoagulant of choice, and in women with OHSS, LMWH is given for at least 3 months following symptom resolution of OHSS ( Tables 3 and 4 ). 20 In less severe OHSS, anticoagulation should follow the guidelines for the management of pregnancy, including LMWH prophylaxis in the co-occurrence of some high-risk thrombophilias that warrant anticoagulation, as outlined in the ASH guidelines. 20 , 54

Just as in women, testosterone treatment in men increases the risk of VTE. 64 – 66 In a case-crossover study of over 39,000 men at an average age of 57 years, the OR for VTE was 2.32 [95% CI, 1.97–2.74] in those with hypogonadism vs 2.02 [95% CI 1.47–2.77] in those without hypogonadism ( Table 1 ). 66 As observed in premenopausal women receiving hormonal contraception and postmenopausal hormone replacement therapy, the time of greatest VTE risk in men receiving testosterone therapy is early, specifically within the first 6 months of therapy. 64 It has been noted in some studies that VTE may recur despite anticoagulation therapy, which may suggest the co-existence of thrombophilia. 65 Among males with Klinefelter syndrome, a chromosomal disorder and the most common cause of primary testicular failure, the use of testosterone replacement therapy is associated with an increased risk of VTE, not infrequently noted in those with underlying thrombophilia. 67 , 68 Osteonecrosis of the femoral head has also been observed in men receiving testosterone therapy, 44 as noted in healthy pregnant women. 42 Despite anecdotal studies of anticoagulation, the optimal management for these conditions remains unknown. 44 , 69 , 70 In women or men receiving transgender hormone therapy who develop acute VTE, standard anticoagulation guidelines apply, including UFH or LWMH for acute thrombosis, followed by 3–6 months of LWMH or DOACs ( Tables 3 and 4 ). 17 – 19 , 71 Whether prophylactic anticoagulation should be continued long term in those who develop VTE while on hormonal therapy, or in those who develop osteonecrosis, 72 of in those with co-existent thrombophilia, 65 remain unanswered questions which will require future studies to guide management.

Pregnancy is associated with a well-described increase in the risk of venous and arterial thromboembolism. 30 Compared to nonpregnant women, the risk of VTE in pregnant women is increased 4–5-fold and then increases to 20-fold in the postpartum period ( Table 1 ). 31 , 32 In the MEGA study (Multiple Environmental and Genetic Assessment of risk factors for first VTE, a population-based case-control study, the odds for VTE was 4.6 [2.7, 7.8] as compared with nonpregnant women. 32 In fact, this study established that the majority of all VTEs in women of reproductive age are pregnancy associated. The risk appears greatest in the third trimester, but during the 6 weeks postpartum, the odds for VTE is 60-fold greater, 60.1 [26.5, 135.9], and thus constitutes the time period of greatest risk for pregnancy-associated thromboembolism. 32 While some thrombotic events may be attributed to decreased venous capacitance and venous outflow secondary to mechanical obstruction of the pregnant uterus 33 or to anatomic changes or vascular injury, 34 this does not explain thrombosis risk associated with pregnancy. In fact, thrombosis risk begins early in the first trimester of pregnancy and is strongly correlated with increasing estradiol and progesterone levels, 35 underscoring the significant role of hormone-induced hypercoagulability throughout pregnancy. 34 For example, the median estradiol level increases from 2.32 ng/mL in the first trimester to 9.00 ng/mL in the second trimester, and 22.6 ng/mL in the third trimester, while median progesterone levels increase from 25.6 ng/mL, to 48.1 ng/mL, to 130 ng/mL, respectively. 36 During pregnancy, there are increasing levels of coagulation factors VII, VIII, X, and von Willebrand factor, fibrinogen and increased thrombin generation. 37 These changes appear to be balanced by downregulation of the fibrinolytic pathway, as evidenced by increases in plasminogen activator inhibitor type 1 and 2, the latter generated by the placenta during the third trimester. 34 , 38 In fact, the increasing hormonal changes and accompanying VTE risk culminate in a period of physiologic hyperfibrinolysis during the first 3 hours postpartum, 39 presumably as a natural protection against thrombosis. The period of relative hypercoagulability and thrombosis risk continues throughout pregnancy and lasts up to 6–8 weeks postpartum, in parallel with the return of factor levels and thrombin generation to baseline. 37 Among risk factors for pregnancy-associated VTE, the most important is a prior pregnancy-associated VTE, with a 3–5-fold increased relative risk. Interestingly, the hormonal-associated VTE risk in pregnancy does not appear to predict long-term VTE risk outside of pregnancy. 40 Other risk factors of VTE during pregnancy include thrombophilia 41 ( Table 5 ), and other comorbid conditions such as obesity, smoking, hypertension, and lupus. 34 Another risk that has been reported in pregnancy particularly in the last trimester is osteonecrosis of the hip. 42 Osteonecrosis, also referred to as avascular necrosis or aseptic necrosis of the femoral head, may also occur with exogenous estrogen supplementation, 43 is attributed to the occurrence of local thrombosis leading to osseous venous outflow obstruction which reduces vascular supply to the hip. 42 , 44 In a case-crossover trial of women with 1 child and 1 hip osteonecrosis, the risk was highest by the last trimester of pregnancy was 14.5 [8.2, 18.3], but progressively fell back to baseline by 27 months postpartum. 42 Because the disorder may be progressive and lead to joint collapse (Ficat stages I or II) and ultimately necrosis and collapse (Ficat III-IV) even at an early age, 44 there has been increasing interest in the role of anticoagulation in ameliorating the course of this disease. In a small case series, anticoagulation in 6 patients with primary osteonecrosis associated with thrombophilia reversed disease progression to collapse (Ficat stage III-IV). 43 , 44 However, whether anticoagulation is protective for secondary causes, ie, pregnancy-associated osteonecrosis, remains a subject of debate and will require future study. LMWH is the agent of choice for treatment of acute VTE and as well for VTE prophylaxis in pregnancy ( Tables 3 and 4 ). 20 VTE prophylaxis is recommended for high-risk groups including unprovoked VTE, hormone-associated VTE, or higher-risk thrombophilias. 20 The duration of anticoagulation is up to 6 weeks postpartum or to complete at least 3 months treatment following an intrapartum VTE. This is generally changed to UFH in the final month of gestation to take advantage of the decreased half-life of UFH which allows for safer spinal or epidural anesthesia. 2 LMWH and UFH are considered safe options for breastfeeding women and are the anticoagulants of choice in the postpartum period. 20

With their range of clinical applications, estrogen-, progestin-, and testosterone-based therapies will continue to have a prominent role for women and men across the reproductive spectrum and beyond. The thrombosis risk associated with these hormonal therapies has been better defined, allowing for more clear risk-benefit conversations with patients considering treatment. As each formulation and method of delivery carry varying degrees of risk, clinicians are faced with a myriad of treatment choices, which will require ongoing research to ensure that VTE risk is clearly defined as new generations of hormonal therapies come to market. Finally, it will be important to determine whether thrombophilia screening at the initiation of exogenous hormone therapy is cost effective, as has been shown in relatives of factor V Leiden carriers initiating oral contraceptives. 96

Within the transgender community, hormone replacement has an increasingly important role in gender transition. Of particular concern in this group is the risk of VTE associated with the use of estrogens in combination with antiandrogens in male-to-female transsexual patients ( Table 1 ). For transsexual individuals, cross-sex hormone administration is a critical aspect of their treatment program. 6 The administration of these agents is deemed safe when formulations and dosages are based on Endocrine Society guidelines. 55 The risk of VTE associated with cross-sex hormone treatment observed in male-to-female transsexual patients however, is increased, 3.2 [1.5, 6.5] Table 1 ). 56 , 57 Similar to the trend seen in hormonal contraception for women, the original formulations contained higher concentrations of estrogens, such as ethinylestradiol 0.1 mg per day which was associated with a 45-fold increase in VTE 58 in the presence of additional known VTE risk factors such as age, surgery, and immobility. 6 For transgender patients, the standard of care has now become transdermal estrogen. In contrast to the lower risk seen in postmenopausal women, transdermal estrogen in premenopausal women is associated with higher plasma levels of estrogen and increased rates of VTE compared to oral preparations; however, it is unclear which group male-to-female patients mirror. 59 Compared to other estrogen preparations, ethinylestradiol also carries a 3-fold increase in cardiovascular mortality for male-to-female patients age 40–65. 60 Estimates of VTE risk in the modern era of estrogen therapy vary by study with incidences ranging from 2% to 6.4% or 21 to 58/10,000 patient use years, the latter 20-fold higher than the general population. 61 The differences in incidence are attributed to varying estrogen doses and plasma targets, and the study sites with higher incidences also reported higher doses of estrogen administered. 61 For women receiving combination hormonal contraception, the risk of VTE is found to be higher during the initiation of treatment and decreases after a year of therapy. This trend, however, is not seen in male-to-female patients receiving estrogen replacement. A cohort study from Kaiser Permanente health system demonstrated a rise in VTE incidence following 2 years of therapy and a continued rise for 5 years in this population. 57 Estrogens are typically combined with hormonal agents that vary by region. In Europe, they are most frequently combined with cyproterone acetate for patients seeking male-to-female transition. This combination is observed to have a small but significant increase in activated protein C resistance in women. 62 With the current level of evidence, the risk conferred by the cyproterone acetate in these male-to-female transgender patients is believed to be low. In the United States the preferred combination utilizes the antiandrogen spironolactone opposed to cyproterone acetate. In a retrospective review of 676 patients receiving this combination in the United States the incidence of VTE was low, 7.8 per 10,000 person-years. 63 With the decrease in dosage, departure from ethinyl estrogen preparations, and more aggressive VTE prophylaxis around surgery, the rates of VTE in male-to-female patients have decreased. However, these patients still remain at high risk for VTE, compared to the general population and women using estrogen-containing oral contraceptives.

The effect of endogenous and exogenous sex hormones, particularly estrogen and progesterone, on daily life in women is immense, from contraception to assisted reproduction, restoration of hormone imbalance, treatment of cancer, promotion of strong bones and a healthy menopause. Yet, despite their importance across the lifespan, a major complication is venous thromboembolism (VTE), which is potentially life-threatening. 1 Increased endogenous production of estrogens and progestins is implicated in the increased thrombotic risk during pregnancy, 2 in polycystic ovary syndrome (PCOS), and in vitro fertilization (IVF). 3 The common thread running in these scenarios is an increase in endogenous hormone levels that tips the hemostatic balance toward hypercoagulability. The use of exogenous estrogen and progesterone have a wide range of clinical applications, from hormonal contraception and hormone replacement to cross-hormone replacement in transgender patients. 1 , 4 – 6 Given the importance of sex hormones to the women’s health, there is great impetus to understand the mechanism of and reduction in their thrombotic risk. The aim of this review is to summarize VTE risk associated with sex hormones across a woman’s reproductive years and beyond, with reference to current VTE management guidelines with heparin, low-molecular-weight heparin (LMWH) and direct acting oral anticoagulants.

Since their introduction in the 1960s, women in the reproductive years have used hormonal contraceptives, not just for pregnancy prevention, but also in the reduction in dysmenorrhea, endometriosis, menstrual migraines, acne, and hirsutism. 7 For the majority of reproductive-age women, the absolute risk conferred by hormonal contraception is small and outweighed by their potential benefits. Yet, a major risk in women using combination hormonal preparations is VTE, 1 with a 4-fold increased risk, RR 3.5 [2.9, 4.3] as compared with nonusers ( Table 1 ). 8 , 9 VTE risk appears to vary based on the formulation, delivery, and dosage of estrogen, as well as the type of progestin used ( Table 2 ). 10 In fact, most of the VTE risk with combined hormonal contraception is related to the dose and type of estrogen used. Historically, ethynylestradiol was the only estrogen component because of its bio-availability and potency following metabolism in the liver. 4 With the recognition, however, that estrogen dose was related to thromboembolism risk, the dose of estrogen was progressively reduced, from 105 μ g of active ethinylestradiol to as low as 15 μ g in third-generation combinations, 5 with subsequent studies confirming reduction in VTE risk with reduction in estrogen dose. VTE risk with formulations containing 30–40 μ g estrogen have been associated with a 3.2-fold risk [2.7, 3.8] or an incidence of 6.2 events per 10,000 exposure years, 11 while for formulations containing less than 20 μ g VTE risk has been less-well defined. 5 Nonoral estrogen administration, including vaginal rings, transdermal patches, and subcutaneous depots, has been associated with minimally higher rates of VTE than oral formulations. 12 In a Danish population-based study, users of the vaginal ring showed a 6.5-fold increased VTE risk and users of the transdermal patch had a 7.9-fold increased risk, 12 observations supported by pharmacokinetic studies demonstrating that plasma estrogen levels were 60% higher in users of transdermal patches than users of oral contraceptives. 13 Of course, calculation of VTE risk in all these studies is also dependent on variables, in addition to estrogen dose, including, for example, the type of study, population-based vs interventional trial, and the method of VTE diagnosis, by clinical vs Doppler or radiographic confirmation. While progestin preparations are considered low risk for VTE, there is evidence the risk varies with the type of progestin formulation. 1 The use of depot-medroxyprogesterone acetate, a long acting progesterone-only contraceptive is associated with a nonsignificant increase in thrombotic risk, odds ratio (OR) 2.2 [0.7, 7.3], consistent with lack of change in surrogate markers of thrombotic risk, including coagulation factor levels and d-dimer. 14 By contrast, the highest VTE risk among progestins has been reported with third-generation formulations, specifically desogestrel, cyproterone acetate, and drospirenone ( Table 2 ). 11 , 15 The method of progestin delivery also influences risk of VTE: a meta-analysis of 8 studies identified a nonsignificant increase in VTE risk for progestin-only oral contraceptives, relative risk (RR) and 95% confidence interval (CI), (RR 0.90, 95% CI 0.57–1.45); progestin intrauterine devices (IUD), RR 0.61, [0.24–1.53]; and subcutaneous progestin implants, RR 1.4 [0.6–3.4]. 16 It is of note that the highest period of risk for VTE is within the first year of use of estrogen or progestin, or at the time of re-initiation of hormone therapy following at least a 1-month break, after which the risk falls and stabilizes thereafter. 7 In most young women, the VTE risk associated with combination contraception is low, and it is recommended that the lowest tolerable dose of a later-generation estrogen be used. For women at higher risk of VTE, such as obesity, previous VTE, or thrombophilia, or for those who do not require estrogen, the lowest risk option is the progestin IUD or, alternatively, other progestin-only formulations. In the event of VTE associated with hormonal contraception use, it is recommended the hormonal agent be stopped, and nonhormonal or progestin-only contraception, such as the progestin IUD, be subsequently considered. For acute VTE management, unfractionated heparin (UFH) or LMWH are typically given, followed by at least 3–6 months of anticoagulation with LWMH or oral direct acting anticoagulant (DOAC) 17 – 19 ( Tables 3 and 4 ). It is recommended that women with a previous hormonal VTE be anticoagulated during subsequent pregnancy and for 6 weeks postpartum, with the agent of choice, LMWH. 20

Hormonal therapy in postmenopausal women has successfully improved quality of life and decreased symptoms associated with menopause. Historically, estrogen formulations were the principle therapy to relieve significant vasomotor symptoms which occur in up to 75% of postmenopausal women. 84 These formulations contain one-fourth to one-fifth of the estrogen levels used in oral contraceptives, but throughout the late 1990s and early 2000s, there was mounting evidence these drugs were associated with increased risk of VTE ( Table 1 ). 85 In fact, the findings of the Women’s Health Initiative (WHI) clinical trial of combination estrogen plus progestin hormone replacement therapy led to a sea change in postmenopausal estrogen supplementation, 3 , 86 with a 50% decline in the use of hormone therapy in the United States. 87 The randomized controlled Heart and Estrogen/progestin Replacement Study trial demonstrated that daily 0.625 mg equine estrogen plus medroxyprogesterone acetate in women with known coronary artery disease increased VTE risk 3-fold compared to pacebo, 88 confirming previous observational studies 87 and the findings in randomized trials, meta-analysis, and population-based case-control studies. 33 , 86 , 88 – 91 By contrast, based on systematic reviews and meta-analysis, nonoral replacement with transdermal and estrogen only formulations did not increase VTE risk. 84 Similar to hormonal contraceptive use in premenopausal women, the time of greatest VTE risk was within the first year of therapy. The type of progestin also impacts hormonal VTE risk in postmenopausal women. The rationale behind the use of progestin in postmenopausal women is to protect uterine tissue from estrogen-induced endometrial hyperplasia and adenocarcinoma. 92 For women who have undergone hysterectomy, single-agent estrogen without progestin is typically prescribed, given its lower risk of VTE compared to combination estrogen and progestin, typically medroxyprogesterone acetate, use in the United States. 93 In Europe, a greater variety of progestin types are utilized, including norpregnane derivatives which, with their higher affinity for the progesterone receptor, have no androgenic, estrogenic, or glucocorticoid activity, and in case-control studies do not increase VTE risk, 7 findings confirmed by the WHI Estrogen-Alone trial. 38 , 87 , 94 , In this randomized placebo-controlled WHI Estrogen-Alone trial, postmenopausal women with prior hysterectomy randomized to conjugated equine estrogens had a lower risk of deep venous thrombosis than placebo, 0.63 [0.41, 0.98]. 94 Despite the alleviation of menopausal vasomotor symptoms, osteoporotic fracture, and reduced colorectal cancer risk, the use of hormonal therapy, whether oral or transdermal, has decreased significantly over the past several decades because of increased VTE risk. For that reason, when hormone replacement therapy is required, the recommended approach is transdermal estrogen for women following hysterectomy, and transdermal estrogen-norpregnane combinations for those with an intact uterus, given the low risk of VTE. In general, a previous history of VTE due to pregnancy, oral contraceptives, unprovoked or associated with thrombophilia is considered a contraindication for estrogen therapy. 95 By contrast, for women with VTE associated with immobility, in the setting of surgery or fracture, transdermal but not oral estrogen therapy have been considered. 95 Occurrence of VTE in the setting of endogenous or exogenous hormone exposure requires anticoagulation as set forth by the American Society of Hematology. 17 , 18 , 20 For acute thrombosis, UFH or LMWH therapy is warranted, followed by 3–6 months of prophylaxis with LMWH or DOACs ( Tables 3 and 4 ). 17 – 19