Sex hormones and stroke: beyond estrogens

Abstract

Stroke risk and poor stroke outcomes in postmenopausal women have usually beeen attributed to decreased levels of estrogen. However, two lines of evidence suggest that this hormone may not be solely responsible for elevated stroke risk in this population. First, the increased risk for CVD and stroke occurs much earlier than menopause at a time when estrogen levels are not yet reduced. Second, estrogen therapy has not successfully reduced stroke risk in all studies. Other sex hormones may therefore also contribute to stroke risk. Prior to menopause, levels of the gonadotrophin Follicle Stimulating Hormone (FSH) are elevated while levels of the gonadal peptide inhibin are lowered, indicating an overall decrease in ovarian reserve. Similarly, reduced estrogen levels at menopause significantly increase the ratio of androgens to estrogens. In view of the evidence that androgens may be unfavorable for CVD and stroke, this elevated ratio of testosterone to estrogen may also contribute to the postmenopause-associated stroke risk. This review synthesizes evidence from different clinical populations including natural menopause, surgical menopause, women on chemotherapy, and preclinical stroke models to dissect the role of ovarian hormones and stroke risk and outcomes.

Menopause, estrogen deficiency and stroke outcomes:

The importance of ovarian hormones as a risk factor for stroke is evident in comparisons of the incidence of female strokes before and after menopause. Premenopausal women have a much lower incidence of stroke as compared to young males, however at the menopause transition (ages 45–54), the incidence of stroke is double that of men^1,2. In tandem with increased risk, stroke outcomes are also worse (reviewed in³). Women account for 60% of stroke-related deaths⁴, even after normalization for age. A Canadian stroke registry study reported that 10% of women stroke patients were discharged to long term care as compared to 5% men⁵, despite the fact that stroke size tends not to be different in males and females⁶. Moreover, 5-yr stroke recurrence is disproportionately higher in females (20%) as compared to males (10%) in the 45–64 age range⁷.

Increased stroke risk and severity among older women led to the hypothesis that the loss of ovarian hormones, principally estrogens, at menopause may be a contributory factor. However, analysis of hormone use and stroke incidence in pre and postmenopausal women does not support this conclusion entirely. For example, a multicenter case-controlled study showed that increased lifetime exposure to estrogen was associated with a lower risk of stroke, supporting the idea that estrogens are benefical⁸. In contrast, a case-control study in Northern California Kaiser Permanente facilities reported no benefit for stroke risk in postmenopausal women who took hormone therapy relative to those not taking hormones⁹. More ominously, the Women’s Health Initiative (WHI) study, which had a signifcant impact on menopause medicine, concluded that hormone use actually increased stroke risk. This randomized, double blind, placebo-controlled multicenter trial compared the risk of myocardial infarction, stroke and dementia in women who consumed daily conjugated equine estrogens (CEE)¹⁰, CEE+progestins¹¹ or placebo. Hormone therapy groups showed an increased risk for stroke; however, subgroup analyses indicated that most of this risk was seen in the older age groups. In the CEE trial, stroke risk was significantly elevated in the 60–69 year old group but not the 50–59 year old group¹⁰. In an observational analysis of postmenopausal women in the Nurse’s Health Study, estrogen and estrogen+progestin use increased the risk of stroke irrespective of the age of the user or time since menopause¹². However, the observational arm of the WHI study showed no increased risk for stroke in the CEE or CEE+progestin arm^13,14. A possible factor in the discrepancy between the WHI trial and the WHI observational study was that the initiation of hormones was much earlier in the latter study. However other health characteristics among this group can also impact stroke risk in conjunction with hormone therapy (HT). In the observational trial (SHOW study) HT users were more likely to be normotensive and lean as compared to non-users in this study¹⁵ which was not the case in the WHI study, where hypertension incidence was similar in CEE users and non-users¹⁰. A similar interaction between HT and hypertension was seen in the Danish Nurses study, where normotensive women who used hormone therapy were not different from controls, while the risk for stroke was elevated among hypertensive women who used hormone therapy¹⁶.

In addition to comorbid conditions, hormone treatment effects are also modified by the timing of treatment. Data from a prospective study of Swedish women showed that stroke risk was significantly decreased in women who initiated hormone treatment prior to menopause¹⁷. In a population-based nested case-control study of 50–69 year old women, HT did not significantly elevate ischemic stroke risk¹⁸, further supporting the idea that HT at ages closer to the menopause may be harmless for stroke. Coronary artery calcification, a surrogate marker of cardiac disease, was reduced by estrogen in the youngest cohort of the WHI study (50–59 years)¹⁹, also signifying that estrogen’s effects can be modulated by the age of the user. Finally, a study of non-users of HT found that stroke-related mortality in women 65 and older was higher in women with higher levels of endogenous estrogen²⁰, implying that elevated levels of hormones in late life, whether exogenous or endogenous, may exert a deleterious effect on stroke. The issue of timing of treatment was directly tested in the Kronos Early Estrogen Replacement study (KEEPS) ²¹ and the Early versus Late Intervention Trial with Estradiol (ELITE) study²². The KEEP study was a prospective, randomized, controlled trial study where the primary outcome measure was cardiovascular risk measured by carotid intima thickness, coronary artery calcium, as well as other ancillary measures. Participants were women who were within 3 years of menopause and received either oral CEE or transdermal 17b-estradiol²¹. The study found that there was no difference in carotid intima thickness (CIMT) in the oral CEE, transdermal 17b-estradiol or placebo-treated groups²³. Moreover, although hormone treatment did not affect cognitive function²⁴, it improved sleep quality and vasomotor symptoms²⁵. These findings suggest that hormone therapy for healthy, early postmenopausal women does not increase cardiac disease indicators. The ELITE trial tested the effect of oral 17b-estradiol treatment (with or without progesterone by vaginal gel) on early (<6 years) and late (>10 years) post-menopausal women. This study showed that progression of CIMT was influenced was influenced by the timing of estradiol treatment. Thus estradiol treatment to the early menopausal group had a lower rate of progression of CIMT as compared to placebo controls while CIMT measures in estradiol treated late post-menopausal group were no different from the placebo group²². The ELITE study therefore suggested a protective role for 17b estradiol for early menopausal females. Neither study examined stroke as an endpoint, but extrapolating from the CVD marker, these studies suggest that estrogens are not deleterious when given to women at the early stage of menopausal. In summary, the evidence linking estrogen therapy and stroke risk, is modified by several intervening variables. The modification by ‘age’ (early or late menopause) suggest that other sex hormones may also influence this association.

1. If not estrogen, then what?

The aging ovary: The normal reproductive ovary secretes hormones under the regulation of two pituitary gonadotropins, follicle-stimulating hormones (FSH) and luteinizing hormone (LH) in a tightly regulated cycle²⁶ (Figure 1). In this section, three types of ovarian secretions, steroids, gonadal peptides, and growth factors, will be reviewed.

Figure 1:

Schematic representation of the pituitary-ovary axis at (A) pre-menopause and (B) menopause. (A) Pituitary gonadotrophin release acts on theca cells to stimulate androgen synthesis, while FSH increases aromatase in granulosa cells, which facilitates conversion of testosterone to estrogen. Estrogen subsequently suppresses FSH. Menopause: Age-related follicular atresia is associated with loss of granulosa cells and theca cells that accumulate as hilar cells. Pituitary secretions elevate androgens in theca cells, which is not converted to estrogens due to granulosa cell loss, thus increasing T:E ratio and FSH due to loss of negative feedback. LH: Luteinizing hormone, FSH: Follicle stimulating hormone, TC: theca cells, GC: granulosa cells, AL: anterior lobe of the pituitary, PL: posterior lobe of the pituitary.

Ovarian steroids are essential for preparing the uterine endometrium for pregnancy²⁷. Specifically, in response to LH stimulation, ovarian theca cells produce the androgens, DHEA, testosterone, and androstenedione, which diffuse into the granulosa cells^28,29. The granulosa cells, in turn, upregulate aromatase in response to FSH to produce 17β-estradiol (E2)³⁰ (Figure 1). After ovulation, the corpus luteum forms, and its theca and granulosa cells increase its production of the other major sex steroid, progesterone³¹.

The ovarian peptide hormones regulate ovarian function and health. Inhibin, activin, and follistatin modulate pituitary gonadotropin release^32,33,34,35. Insulin-like growth factors (IGF) −1 & −2 are produced by ovarian granulosa cells and are involved in follicle development^{36,37,38,39,26}. FSH also regulates IGF binding proteins and stimulates ovarian IGF-1 synthesis⁴⁰. Antimullerian hormone (AMH) is produced by pre-antral and early antral follicles and reflect the approximate size of the primordial follicle pool and maybe the best biochemical marker of ovarian function⁴¹. AMH levels gradually decline with age and loss of the primordial follicle pool. At menopause, AMH is undetectable⁴².

After menopause, the ovary becomes unresponsive to pituitary gonadotropins, and ovarian hormone production declines⁴³. Estrone (E1), which is produced mostly by the aromatization of androstenedione in fat, replaces 17β-estradiol (E2) as the dominant circulating estrogen^44,45,46. However, estrone levels are at lower quantities than premenopausal E2^47,48. Ovarian testosterone production is preserved and regulated by the gonadotropin LH^49,50. The growth factors and gonadal peptides are not produced in appreciable amounts in postmenstrual women^51,49. In non-human primates, the data is contradictory with studies showing ovarian hormone decline with age, and other showing that menstruation occurs in late life⁵². Rodents, like humans, also cease cyclic hormones expression during reproductive senescence. Rats and mice both have 4–5 day estrus cycles, and both species will display age-related changes in cycle length and pattern, although these changes may occur at different chronological ages depending on the species and strain. Thus, Fisher rats do not become acyclic until 18+ months of age⁵³, while Sprague Dawley rats are acyclic at 10–12 months of age^54,55. Among mice strain differences have not been well studied, however, C57/B6 mice, one the most commonly used transgenic and wild type model, are acyclic at 13–16 months of age⁵⁶. With age, rodents display an acyclic pattern of persistent estrus where estradiol levels are still measurable, and FSH levels are low⁵⁷. This is followed by persistent diestrus, where estradiol levels are undetectable and FSH levels are elevated⁵⁸. In rodents, the role of estrogen has been thoroughly studied, however, the actions of other ovarian and pituitary hormones is poorly understood (see Table 1 for summary).

Table 1.

Sex differences in stroke outcomes	Potential mechanism
Young female rats have smaller infarcts than male rats and this sex difference is eradicated by bilateral ovariectomy161, 186	Estrogen deficiency worsens stroke outcomes
Estrogen treatment to ovariectomized young females is neuroprotective164, 167, 168, 170

Age differences in stroke outcomes	Potential mechanism	Future direction
Worse stroke outcomes in middle-aged/aged females as compared to young females or to age-matched males	Toxicity due to low estrogen levels	Use estrogen levels that suppress FSH or improve T:E ratio, instead of matching pre-senescent levels of estrogen
	-Estrogen treatment to ovariectomized middle-aged or aged females is toxic187, 188
	Toxicity due to elevated levels of FSH	Test stroke outcomes in animals with FSH over-abundance or FSH receptor k/o.
	FSH receptor knockout in mice: Promote bone health75–77
	Estrogen + FSH treatment: No atheroprotection68
	Toxicity due to increased T:E ratio	Decreasing free testosterone during postmenopausal transition could improve stroke outcome
	Testosterone exacerbates stroke infarction Castrated males have better stroke outcomes than gonadally-intact males183–185

FSH as a risk factor for CVD and stroke:

Follicle Stimulating Hormone (FSH) stimulates maturation of follicles and estrogen synthesis, and free estrogen, in a negative feedback loop, inhibits FSH secretion from the pituitary. However, serum FSH starts rising above the normal level before menopause, when the level of estrogen is still normal⁵⁹. Unlike estrogen, FSH has an unfavorable effect on lipid profiles. FSH promotes lipogenesis and fat storage, which is reduced in the FSH receptor knockout animals^60,61. Pre-menopausal women with serum FSH >7 IU/l have significant elevation of serum total cholesterol compared to pre-menopausal women with serum FSH < 7 IU/l⁶². Adipose tissue, which is the main repository for cholesterol, secretes pro-inflammatory molecules such as tumor necrosis factor-alpha (TNF-a) and Interleukin-6 (IL-6), which are associated with development of atherosclerotic plaques and insulin resistance^63,64. Not surprisingly, in postmenopausal women, FSH levels also correlate well with vascular inflammation^65,66. In the SWAN study, low FSH was associated with low intima media thickness (a measure of cardiac health), as compared to medium or high FSH levels⁶⁷. Higher levels of FSH and Vascular Cell Adhesion Molecule (VCAM)1, an adhesion molecule associated with atherosclerotic lesions, are found in blood samples from postmenopausal women as compared to premenopausal women, with a positive correlation between these molecules⁶⁸. In vitro, FSH is shown to elevate VCAM1 in endothelial cells and increase adhesion of human monocytes to endothelial cells⁶⁸. Interestingly, in a 22-site population-based study of postmenopausal women in East China, high levels of FSH were associated with a low risk of atherosclerotic cardiovascular disease, and low obesity⁶⁹. In fact two recent studies report that FSH levels are better predictor of metabolic disease in postmenopausal women than CRP (C Reactive Protein), adiponectin and leptin^70,71. Thus, the effect of FSH effect is non-linear, and may affect risk factors for cardiovascular disease differently from vascular inflammation.

In addition to the cardiovascular health, FSH also negatively impacts bone health. Elevated level of circulating FSH in the premenopausal period (when estrogen levels are normal) is associated with post-menopausal bone turnover; especially bone resorption⁷². When estrogen level is low, such as cases of amenorrhea, a high level of FSH is strongly correlated with decreasing bone density⁷³. Hormone therapy is generally shown to prevent bone loss in estrogen-deficient women, but a subgroup of postmenopausal women still lose bone mass despite HT. A retrospective study showed that serum FSH level is a good predictor of bone mineral density in patients with hormonal replacement therapy⁷⁴. These finding are supported by animal studies which showed that mice with FSH receptors knock out or simple hypophysectomy can preserve bone density despite severe hypogonadism^75–77.

1.2. Androgens:

The ratio of estrogen and androgen is also a critical factor for physiologic homeostasis. Estradiol, the most active estrogen, is primarily produced by ovary, whereas testosterone is produced by the ovary (25%), the adrenal gland (25%) and by peripheral conversion of androstenedione and dehydroepiandrosterone (produced by ovarian stroma and adrenal gland) (50%)⁷⁸. After menopause, the source of circulating testosterone is: 50% by ovaries, 10% by adrenal gland and 40% by peripheral conversion of androstenedione and dehydroepiandrosterone⁷⁹. During the postmenopause period, the ovary increases secretion of testosterone possibly via stimulation by elevated LH, and this secretion sometimes span up to 10 years after menopause^80–82. As a consequence, estrogen levels decline steeply after menopause^79,83 whereas testosterone level remain more or less unchanged, leading to a state of relative androgen excess^84–86. Like FSH, excess androgen also has unfavorable effects on lipid profiles. It decreases HDL level and increases triglyceride, LDL and total cholesterol^87–89. Testosterone is also positively correlated with insulin resistance and type-2 diabetes in elderly population⁹⁰. Thus one possibility is that estrogen therapy may be more effective if the imbalance between androgen and estrogen is normalized, especially during the transition during the pre- and post- menopause stage.

1.3. Sex hormone binding globulin (SHBG):

Studies have stressed that along with androgen and estrogen, sex hormone binding globulin (SHBG) be included in evaluating risk factors for cardiovascular diseases and mortality in postmenopausal women. SHBG levels tend to decrease across the menopause transition^91,92. Low SHBG significantly correlates with the incidence of non-insulin-dependent diabetes mellitus (NIDDM), and stroke^93–95, and high level of SHBG in postmenopaual women is strongly associated with decreased risk of type-2 diabetes⁹⁶. Similar to high level of FSH and androgen, low SHBG is associated with increased triglyceride⁹⁷, decreased HDL-C⁹⁸ and, in elderly men, it is associated with coronary heart disease (CHD) mortality⁹⁹. A high level of free testosterone and low level of SHBG is also associated with elevated triglyceride and low level of HDL¹⁰⁰. FAI (Free Androgen Index: Free Testosterone level / SHBG) increase during the menopausal transition is also strong risk factor for metabolic syndromes occurring around the menopausal phase^86,101. Thus, in addition to low estrogen levels SHBG may also modify cardiovascular pathologies.

Early menopause:

Two populations can inform the debate on early menopause and stroke risk:

2.1. Chemotherapy for breast cancer:

More than 250,000 women will be diagnosed with breast cancer every year, and women with ER+ tumors¹⁰² will be prescribed adjuvant therapy involving endocrine agents. Tamoxifen (TAM), a selective estrogen receptor modulator (SERM) is the most prescribed therapy for premenopausal breast cancer patients. Alternately, women may be prescribed aromatase inhibitors (AI) which block estrogen synthesis. The third-generation AIs such as anastrozole, exemestane and letrozole have largely replaced tamoxifen as the preferred treatment for HR+ breast cancer in postmenopausal women¹⁰³ and are more effective in women with ER+/PR-tumors. In premenopausal females, AIs are only used if TAM treatment fails¹⁰⁴, and then usually in conjunction with ovarian suppression drugs such as GnRH agonists¹⁰⁴. Despite their overall effectiveness, however, long term usage of these treatments reveal moderate to severe side effects.

Both drugs are shown to increase disease-free survival, and to reduce cancer-related mortality. At the same time, cardiovascular disease has emerged as the single greatest non-cancer cause of death, accounting for approximately 35% of non-breast cancer mortality for survivors 50 years of age and older. The accelerated menopause phenotype is well recognized in this population, including changes in bone, uterine and cardiovascular health. Neurologic changes due to breast cancer therapies are only recently recognized and only infrequently included in the overall assessment of breast cancer survivors. A recent paper¹⁰⁵ clearly outlines this gap in our knowledge of estrogen signaling on neurologic disease such as AD. A similar case may be made for cerebrovascular stroke, since TAM and LTZ causes menopause-like phenotype and effectively places women into a stroke-prone demographic, where outcomes are very poor.

Tamoxifen:

Initially a failed contraceptive compound, tamoxifen (TAM) or ICI146,474 was successfully repurposed as adjuvant breast cancer therapy due to its antagonist actions on estrogen receptors (ER) in this tissue¹⁰⁶. TAM is metabolized to 4-hydoxytamoxifen and can exert both antagonist and agonist actions at the estrogen receptor, depending on the cis/trans conformation of its metabolites¹⁰⁷, and/or the type of steroid receptor coactivators present in the cell¹⁰⁸. Thus, while its action on breast tumor cells is antagonistic, TAM is an agonist on bone tissue and the uterus. Studies have repeatedly shown that TAM use increases the risk for endometrial cancer^109,110,111, pulmonary emboli (ATLAS study)¹¹⁰, and increased mortality due to ischemic heart disease and stroke (NSABP B-14) although patient events were low in the latter¹¹¹. Neural effects of these endocrine therapies are also noted, specifically in cognitive function. In a Dutch study, TAM treatment resulted in significantly lower scores on cognitive performance, especially verbal, memory and executive functioning¹¹². Patients treated for 5–6 months with either TAM or anestrozole (a related AI) showed significant cognitive decline from baseline scores¹¹³, although most studies show that the two classes of drugs themselves do not differ in their cognitive effects (reviewed in¹¹⁴). In addition to tissue-specific effects, TAM is also anti-angiogenic, preventing angiogenesis and endothelial tube formation in vitro¹¹⁵, reducing VEGF-induced angiogenesis in vivo in a matrigel preparation¹¹⁶ as well as in the uterus¹¹⁷, indicating the potential for widespread inhibition of angiogenic signaling in multiple organs.

Letrozole:

Letrozole, a 3^rd generation aromatase inhibitor, effectively inhibits estrogen synthesis throughout the body, and is a highly effective adjuvant therapy for post-menopausal patients. LTZ is a potent non-steroidal aromatase inhibitor, both in vivo and in vitro¹¹⁸. However, LTZ results in significant side effects such arthralgia and higher grade cardiac events¹¹⁹. Unlike, TAM which has tissue-specific effects, LZT inhibits estrogen synthesis in all tissues. As a result, it has significant side effects that resemble menopausal symptoms including vaginal dryness, hot flashes, loss of libido, musculoskeletal pain and loss of bone mineral density. Joint pain or arthralgia which is associated with increased levels of inflammatory cytokines, occurs in 50% of women prescribed AI, leading to 20% non-compliance with an otherwise effective drug¹²⁰. Women on LTZ have elevated levels of inflammatory mediators including CRP, eotaxin, MCP-1 (Monocyte Chemoattractant Protein-1)¹²¹ and in rodents, a three week course of LTZ causes ovarian cysts with elevation of CRP and oxidative stress¹²². Similar to TAM, LTZ exerts suppressive effects on angiogenesis in ovarian tissue¹²³. By inhibiting estrogen synthesis and signaling, both LTZ and TAM have the potential to impair stroke recovery.

Among breast cancer survivors, cardiovascular disease (CVD) accounts for approximately 35% of non-breast cancer mortality for survivors 50 years of age and older, making this the single largest non-cancerous cause of death. Cancer is commonly seen in stroke patients and a recent study reported that the rates of hospitalization for stroke among cancer patients (urogenital, breast, prostate being the most frequent) were significantly higher than non-cancer patients in 1997, and remained high in 2006, while hospitalization rates actually fell among non-cancer patients for stroke during the same time frame¹²⁴. Meta-analyses of breast cancer treatment trials showed that tamoxifen is associated with an increased risk of stroke^125–127. Compared to AI, thromoboembolic events and transient ischemic attacks (TIAs) are more common with TAM treatment¹²⁸.

Unlike TAM, LTZ is more likely to lead to elevated levels of FSH¹²⁹. Interestingly, this group is also more likely to suffer bone fractures compared to TAM¹³⁰, although estrogens are equally reduced in both these treatments¹³¹. Indirectly, these data support the idea that elevated FSH levels, occuring as a result of estrogen depletion, is likely to be deletrious to traditional targets of estrogen. Bone loss is reported to occur 2–3 years prior to the final menstrual period (FMP)¹³², a time at which estrogens are not yet low but FSH levels are significantly elevated¹³³.

Far less, however, is known about the effects of breast cancer hormone therapy on stroke recovery. A retrospective analysis of stroke among cancer patients (urogenital, breast and gastrointestinal, being the most frequent) and controls found that cancer patients had a poorer neurological condition at discharge and a trend towards a longer stay in the stroke unit¹³⁴, both indicative of worse stroke outcomes. The Bergen NORSTROKE study found that ischemic stroke was more prevalent in cancer patients, and the median NIHSS score (an indicator of stroke severity) was significantly higher in cancer patients than non-cancer patients¹³⁵. Given the mixed cancer population in these study, and the lack of information on endocrine therapy use in this group, the recovery from stroke in patients with TAM or LTZ treatment remains an important gap in the health literature. A report from the Swedish National Hospital Discharge Registry combined with the Swedish Cause of Death Registry showed that stroke incidence was increased during the active phase of TAM treatment and reduced after the active period of treatment. Moreover, mortality from stroke also increased during the active drug period and fell during the post-treatment period¹³⁶, indicating that both the risk and severity of stroke is affected by TAM treatment. Currently no studies are available on the effects of LTZ on stroke recovery. Long term TAM or LTZ therapy is understudied in the context of stroke and is poorly studied in the preclinial literature. This group of patients show premature menopause, where both FSH levels are likely to be elevated (due to estrogen suppression) as well as an increased ratio of testosterone to estrogen.

2.2. Premature menopause resulting from Bilateral Salpingo-Oophorectomy (BSO):

Bilateral salpingo oophorectomy is a prophylactic surgery for women positive for BRCA1 and BRCA2 mutations to prevent ovarian or fallopian cancer, after their childbearing age. However the majority of cases involve benign disease. BSO is also usually performed along with hysterectomy in order to avoid ovarian pathologies or adnexal pain from postsurgical adhesions. In 1984, 1 out of 8 US women were reported to have undergone oophorectomy prior to natural menopause¹³⁷, however oophorectomy surgeries have decreased overall nationwide since that time. A recent study shows that the rate of elective bilateral salpingo-oophorectomy was 7.8 per 10,000 in 1998, which increased to 9.0 per 10,000 in 2001 and then fell to 7.4 per 10,000 in 2006¹³⁸. Due to the increased prevalence of outpatient BSO surgeries, these numbers may be an underestimate since most databases reflect inpatient surgeries¹³⁹.

Oophorectomy reduces estrogens and testosterone levels and causes a corresponding rise in FSH levels¹⁴⁰. In a community-dwellers study (Ranch Bernardo), testosterone levels were significantly reduced in women with BSO when compared to age-matched women with no surgery and women who undergo hysterectomy with preservation of ovaries¹⁴¹. In contrast, FSH levels are significantly elevated in this population. Typically, FSH levels of 30 IU/L with one year without menses is indicative of menopause. Using a surgically intact referent group of women, women over 40years of age with hysterectomy and unilateral oophorectomy were 2.49× more likely to exceed a 40 IU/L criteria and 19.17X more likely to exceed this criteria in women under 40 years of age¹⁴².

While BSO reduces ovarian cancer-related mortality in women inherited with BRCA1 and BRCA2 mutations¹⁴³, it may do more harm than good for women¹⁴⁴, including increased non-cancer mortality^{145, 146}. Among the earliest studies on this population were short term (3–6 months) assessments of cognition. Oophorectomy decreased scores on tests of cognition and recall and the MMSE, while estrogen therapy maintained scores at presurgical levels^147–149. Perhaps the most definitive studies of health risks for BSO come from longitudinal studies on a well characterized cohort in Olmstead County, MN, called the Mayo Clinic Cohort Study of Oophorectomy and Aging. Women in this cohort showed significant neurological and cardiovascular health risks¹⁴⁵. In this cohort, bilateral oophorectomy performed at a younger age, in the absence of estrogen therapy, leads to higher risk for neurological diseases such as Parkinson (Hazard Ratio (HR): 1.8), Dementia (HR=1.7), and depression (HR=1.54)^{144,150–152}. Similarly, an increased risk of cognitive impairment was also found after unilateral or bilateral oophorectomy before menopause¹⁵³. Estrogen therapy to women with BSO after 49 years of age showed no increased risk for dementia¹⁴⁴.

In addition to neurological disorders, bilateral oophorectomy also led to cardiovascular disorders in women. Compared to pre-menopausal women, oophorectomy in the absence of estrogen therapy in age-matched women increased the risk for MI by more than 2-fold^144,154. Similarly, an observational study on 29,380 women participating in The Nurses’ Health Study (NHS) with age range of 30–55 also found that compared to women with ovarian conservation, oophorectomy without estrogen therapy increased the risk for MI¹⁵⁵. BSO after the age of 50 increases the risk for stroke, which is attenuated by estrogen therapy¹⁵⁶.

Increase in serum lipids, reduced carotid artery blood flow, and increased atherosclerosis due to the reduced circulating level of estrogen in the oophorectomized women could be the reason for these cardiovascular conditions. BSO before 45 years, with no estrogen therapy, increased all-cause mortality^{143,146,155,157}. In many of these studies, HT reduced the risk for stroke. While this may indicate that estrogen is a mediator of stroke effects¹⁵⁸, it does not preclude the involvement of FSH, since estrogen therapy is known to suppress FSH¹⁵⁹.

3. Preclinical models:

Lessons learnt from Preclinical Studies

While it is not practical to study stroke risk in preclinical models, a significant literature is available on the role of sex hormones in stroke recovery. Ischemic stroke is usually caused by mechanical or biochemical occlusion of the lumen of a major brain vessel, typically the middle cerebral artery (MCA) which results in a corticostriatal infarct, or bilateral carotid artery occlusion that results in hippocampal cell death, among others¹⁶⁰. Stroke-induced neural damage is usually assessed by a variety of measures such as infarct volume (extent of dead tissue), changes in vascularity (microvessel density, tortuosity, length), changes in blood brain barrier permeability, central and peripheral inflammation, and stroke impairment and recovery is measured by short term and long-term deficits in sensory motor function, cognition, depressive-like behaviors, among others.

In rodents, ischemic stroke results in a smaller infarct and better cerebral blood flow in young females compared to age-matched, normoglycemic¹⁶¹ or diabetic¹⁶² males. This sex difference is eradicated when females are bilaterally ovariectomized, supporting the idea that gonadal hormones may underlie these sex differences. This idea received further support from studies showing that the extent of ischemic damage was inversely related to circulating levels of estrogen¹⁶³. Over the last 20 years, studies have overwhelmingly shown that estrogen treatment to ovariectomized female rats or mice is neuroprotective, an umbrella term typically referring to reduced infarct volume, reduced inflammation and improved motor (behavioral) performance^164–167. Thus, replacement with 17β-estradiol or its inactive stereoisomer 17α estradiol¹⁶⁸ as well as the conjugate equine estrogen preparations¹⁶⁹, all reduced infarct volume in female animals. Exogenous estrogen replacement was shown to be neuroprotective when given prior¹⁷⁰ or subsequent to the injury^171,172.

However, the effects of estrogens on stroke are not always neuroprotective. In some cases, this has been attributed to the age of the animal or the type of ischemic injury.

Type of injury: Most studies use a transient focal ischemia, where blood flow to the middle cerebral artery (MCAo) is disrupted and then reinstated. In this model, estradiol is typically shown to improve stroke infarction. However, in permanent ischemia models, several studies have shown that estrogen increased infarct volume^173–175.

Hormone effects are more complicated in the context of aging. In general, studies agree that older female rats/mice have worse outcomes as compared to young females. In fact, older females have larger infarct volumes than age matched males¹⁷⁶, indicating a virtual reversal of the sex difference seen in young animals. This loss of ‘female advantage’ appears to be related to ovarian function, such that acyclic middle-aged female rats also display significantly larger infarct volumes as compared to young females¹⁶⁷. These studies used reproductive senescent rats, defined as animals with multiple previous successful pregnancies, current reproductive failure, and vaginal cytology indicative of constant diestrus (a low estrogen state). In this group, serum levels of estradiol are low and FSH levels are elevated, consistent with a ‘menopausal’ pattern⁵⁸. Thus, the large infarcts seen in this group compared to young normally cycling females is consistent with the hypothesis that ovarian aging impairs stroke recovery. Moreover estradiol treatment for 2–4 weeks, which decreased infarct volumes^167,177 and reduced sensory motor impairment in young females, paradoxically, increased infarct volume in reproductive senescent females¹⁶⁷.

The anomalous effects of estrogens on stroke outcomes in animal models is reminiscent of the paradoxical effects of this hormone in clinical studies. We proposed (above) that some of these paradoxical effects of estrogen may be related to the altered endocrine environment in aging such as testosterone levels or other sex hormones (FSH). Preclinical studies could be informative on the role of these hormones, especially in the context of aging, however, they have not been exploited. Most studies use a 2-group approach, comparing ovariectomized (OVX) vehicle-treated animals with OVX+estrogen-treated animals, or a 3-group approach comparing gonadally intact, OVX and OVX+estradiol treated animals. Few studies have incorporated an estrogen receptor antagonist, to define the locus of estrogen action. One study that used the pan-estrogen receptor antagonist, ICI182,780, noted that this drug increased striatal (but not cortical) infarction¹⁷⁸ presumably by blocking endogenous estrogen signaling. However, no differences were reported in infarct volume between the wild type and ER-alpha knock out (ERKO) mouse, suggesting that estrogen may act via receptor independent process¹⁷⁹. Conversely, exogenous estradiol is reported to improve infarct volume in the OVX wild type but not in the OVX ERKO mouse¹⁸⁰ or in neuron-specific ER knock out mouse¹⁸¹. These studies did not report whether infarct volume in the untreated OVX mouse is similar to WT, thus it is not clear if the baseline in these models is altered. A third strategy involves the use of the aromatase knock-out or ARKO, where all endogenous estrogen synthesis is inhibited. In this model, MCAo results in a larger infarct volume in the ARKO mouse as compared WT or the ovariectomized WT mouse, and infarct volume is reduced by estradiol treatment¹⁸². Moreover, WT animals pretreated with the aromatase inhibitor fadrozole for 1 week also showed worse infarct volumes after MCAo¹⁸²

The OVX model, which shares similarities with the BSO population, may also result in elevated FSH. Thus, an alternate explanation for the effects of estrogen on stroke outcomes in the OVX model is that this treatment improves stroke outcomes by suppressing FSH. Few studies have tested if estrogen treatment to OVX rats (or mice) reduces FSH levels and whether co-treatment with FSH negates the effects of estradiol. In a study of aortic atherosclerotic lesions in the ApoE−/−, OVX mice showed increased lesions which were reduced by estradiol. However, co-treatment of estradiol and FSH abolished the protective effect of estradiol⁶⁸, suggesting that the atheroprotective effect of estradiol was driven by its negative regulation of FSH. No such studies have been performed for stroke studies.

In the case of ovary-intact populations, there is some concordance of the human and preclinical data, such that stroke risk is elevated in older postmenopausal women and stroke outcomes are worse in reproductive senescent rats. In these groups, toxicity of elevated FSH levels may be a concern, as also the increased testosterone to estrogen ratio may be a probable cause for worse stroke outcomes. In rodent models, testosterone has been shown to exacerbate infarction. Thus, castrate males display lower infarct volumes as compared to gonadal-intact males and testosterone replaced castrate males^183,184,185. Thus, larger infarct sizes in reproductive senescent females may also result from elevated testosterone levels. A corollary to this hypothesis is that estrogen treatment to this group is toxic or ineffective because it fails to suppress elevated FSH levels or fails to restore homeostatic levels of testosterone and estrogen. It also raises the intriguing idea that ovariectomy to this group may actually improve stroke outcomes as compared to ovary intact females. These alternate strategies have not been tested thus far.

Conclusions:

Overall, the evidence linking estrogen loss to increased stroke risk is fairly strong, however, the data on estrogen therapy and stroke risk is more ambiguous. In reviewing other cohorts of early menopause (surgical or through chemotherapy) suggests that other sex hormones may also be modify this risk. In the preclinical literature, the variability in stroke outcomes based on age, dose, type of injury may be better explained by considering a global change in ovarian hormones and gonadotrophins.

Highlights.

• Loss of ovarian hormones increases the risk for ischemic stroke in women, and increases stroke severity in animal models

• Estrogen treatment, however, does not consistently reduce stroke risk in women, or stroke outcomes in animal studies.

• The increase in gonadotropins and altered ratio of androgens to estrogens after menopause may be important modifiers of stroke risk and stroke outcomes