Advances in Stroke: Women and Unique Risk Factors

Abstract

There has been a plethora of studies focused on female-specific risk factors and sex differences in stroke published in the past year. This article summarizes several of those novel studies which provide new knowledge about the field of stroke in women. The Nurses’ Health Study II provides new data on the association between laparoscopically-proven endometriosis and future stroke, accounting for the mediation effects of hysterectomy and oophorectomy. In a cohort of women from China, the relationship between hysterectomy, oophorectomy and stroke is further clarified, accounting for the age at which the procedure is performed. The UK Biobank study provides new information on the relationship between oral contraceptive and hormone replacement therapy and stroke, with analytical techniques that focus on the timing of events related to duration of exposure. Finally, two new meta-analyses address the question of whether sex differences exist in the presentation of stroke symptoms.

The relationship between endometriosis and stroke risk from the Nurses’ Health Study II

Endometriosis is a complex condition defined as the existence of extra-uterine endometrium-like tissue.¹ This condition is estimated to occur in 10% of reproductive-age women,² but the true prevalence is difficult to ascertain because it requires surgical (laparascopic) visualization. The pathophysiology of endometriosis involves an inflammatory process, including the infiltration of proinflammatory mediators into the peritoneal microenvironment, which then likely facilitates the clinical symptoms and signs, as well as ovarian dysfunction.³ Endothelial dysfunction and inflammatory biomarkers⁴ such as cytokines have also been associated with endometriosis and therefore this condition has been linked to an increased risk of cardiovascular disease,⁵ including stroke.^6,7 The most recent investigation of the relationship between endometriosis and stroke comes from the Nurses’ Health Study II, a cohort of participants with nearly 30 years of follow-up. Participants with self-reported endometriosis had this diagnosis confirmed based on laparoscopic findings recorded in their medical records. In their analyses of the relationship between endometriosis and stroke in this cohort, the investigators included an important set of potential mediators, including hypertension or hypercholesterolemia, age at menopause (<45 years old vs >45 years old), hysterectomy, oophorectomy, and hormone therapy use, all covariates that are linked to stroke risk. The age-adjusted hazard ratio (HR) for the presence of endometriosis on incident stroke was 1.49 (95% CI 1.23–1.80), and the adjusted HR was 1.34 (95% CI 1.10–1.62), accounting for alcohol use, BMI at age 18, current BMI, age at menarche, menstrual cycle pattern in adolescence, current menstrual cycle pattern in adulthood, parity, oral contraceptive use, smoking, Alternative Health Eating Index (AHEI) diet score, physical activity race, income, family history of MI, family history of stroke, aspirin use, and NSAID use. Importantly, the mediation analysis showed that 39% (95% CI 14.3–71.0) of the relationship between endometriosis and stroke was a result of the hysterectomy and oophorectomy history, with smaller contributions from hormone therapy use, age at menopause, hypertension, and hypercholesterolemia.⁸ Further studies are needed to establish the relationship of endometriosis and the consequences of treatment (medical vs surgical) on stroke risk, but in the meantime, obtaining a thorough reproductive and gynecologic history from women at risk for stroke should be incorporated into prevention practice.

Does hysterectomy alone or hysterectomy plus bilateral oophorectomy increase the risk of stroke in women?

Surgical removal of the uterus (hysterectomy) is the second most common surgery performed on women in the U.S. and is most commonly done in women who are premenopausal (aged 40–44 years).⁹ The relationship between hysterectomy and bilateral oophorectomy on stroke risk in women is uncertain. A prior meta-analysis showed an increased risk of any stroke with oophorectomy, but no clear risk with hysterectomy alone,¹⁰ whereas the UK Biobank study suggested an increased risk of cardiovascular disease but not stroke.¹¹ Although hysterectomy alone does not involve removal of the ovaries, the hysterectomy procedure may disrupt uterine collateral blood flow to the ovaries or have an inhibitory impact on pituitary follicular stimulating hormone (FSH) secretions ultimately leading to early onset of ovarian failure.^12,13 refs.

A recent analysis from the China Kadoorie Biobank study of nearly 300,000 women investigated the relationship between hysterectomy alone (HA; n=8,478) and hysterectomy with bilateral oophorectomy (HBO; n=1360) performed in premenopausal women (mean age at surgery about 43 years) with onset of cardiovascular diseases, including ischemic stroke.¹⁴ In adjusted models, there was no association with either procedure with hemorrhagic stroke, but HA was associated with a 6% increased risk of ischemic stroke (adjusted HR 1.06; 1.01–1.10) and HBO with a 19% increased risk (aHR 1.19; 1.09–1.29). There were also increased risks of total cardiovascular disease and ischemic heart disease of similar magnitudes to ischemic stroke. The risk varied with age at surgery, with a higher ischemic stroke risk in women under age 48 with HA (HR 1.08; 1.02–1.13) and HBO (HR 1.24; 1.12–1.38) but no significant risk in women over age 48. Although this cohort was limited to Chinese women only and should be replicated in a more diverse population, the findings suggest that women who have undergone either HA or HBO, especially if performed at the premenopausal phase, should be screened early for cardiovascular risk factors and treated appropriately.

Is the timing of stroke events after starting oral contraceptives and hormone replacement therapy important?

Meta-analyses examining the relationship between oral contraceptive (OC) use and ischemic stroke reported about a 1.7-fold increased risk of stroke with OC use (RR 1.7; 95% CI 1.5.1.9), with higher relative risks with increasing doses of estrogen.¹⁵ In women with cardiovascular risk factors (hypertension, smoking, obesity, and hyperlipidemia) OC use accentuates the risk.¹⁶ Post-menopausal hormone replacement therapy (HRT) is also associated with an increased stroke risk (RR 1.14; 95% CI 1.04–1.25), with higher risks observed in the setting of later use after menopause and underlying cardiovascular risk.^17,18 The stroke risk associated with HRT appears to be higher for initiation at older ages (farther from menopause), however, given that the use of OCs is typically self-reported, it is not known whether the stroke risks vary with time since initiation. This was the aim of a recently published analysis of the UK Biobank study, a cohort of over 500,000 women ages 37 to 73 years enrolled between 2006 and 2010 at primary care practices in the United Kingdom.¹⁹ The analysis assigned hormone exposure (either OC or HRT) as a time-varying variable, and an observational cohort exposure/non-exposure grouping analogous to the intention-to-treat (ITT) principle used in randomized controlled trials. The authors also performed state-changing analyses which occurred when participants began taking hormones, after the first year of use, and when users reported discontinuation of hormones.

The investigators found that for OC use, there was no increased risk of stroke in the ITT analyses in users vs non-users. However, the time-varying analysis showed an increased hazard of any stroke in the first year of OC use (HR 2.49; 95% CI 1.44–4.30), with no difference in stroke risk between users and nonusers during the remaining years of use and after discontinuation.¹⁹

With HRT, the hazard of any stroke was increased in the ITT analysis (HR 1.22; 1.11–1.35), with the subtype of subarachnoid hemorrhage driving the association (HR 1.33; 1.04–1.71) since there was no increased risk in ischemic stroke or intracerebral hemorrhage. The time-varying analysis, however, showed that the first year of HRT use was associated with a significantly increased hazard of any stroke (HR 2.12; 1.66–2.70), ischemic stroke (HR 1.93; 1.05–3.57), and subarachnoid hemorrhage (HR 2.17; 1.25–3.78). After discontinuation, women who had used HRT continued to have an elevated risk of any stroke (HR 1.16; 1.02–1.32).¹⁹ Although the increased risk of ischemic stroke within one year of initiation of HRT is most likely related to its procoagulant effect, the mechanism for increased risk of SAH is much less clear, and deserves further study, especially given that these models were adjusted for the major risk factors for SAH—smoking and hypertension. Limitations of this study are that the cohort was comprised of only white women who were relatively healthy at enrollment, OC and HRT use are subject to recall bias, and the formulations of both exposures were unknown.¹⁹ However, the analyses have provided some very important contemporary evidence suggesting that, past the first year of use, OCs are likely to confer a low, if any, risk of stroke, but HRT should be cautiously prescribed in women with known stroke risk factors, regardless of menopausal status, and probably avoided in women with known unruptured cerebral aneurysms.

Do women have stroke symptoms that differ from men at presentation?

Many previous studies have reported on potential sex and gender differences in stroke presentation, with variable findings. This year, two new meta-analyses were published with the goal of quantifying the frequency of different presenting symptoms in women and men with stroke or transient ischemic attack (TIA).^20,21 Both studies found that women were more likely than men to present with altered level of consciousness, mental status changes and headache. The meta-analysis by Shajahan et al also identified motor and speech deficits as the most frequent presenting symptoms in both sexes, with no differences between women and men. However, other findings varied between the two meta-analyses; for example, dysarthria was more frequent and paresis less frequent in women in the analysis by Ali et al, but the opposite was found in the analysis by Shajahan et al. Differences in study inclusion criteria likely explain some of the discrepant findings, but the authors also emphasize that a lack of consistency in the collection and classification of symptoms in existing databases hampers our ability to evaluate potential sex differences in presentation and make a case for the use of standardized measures for data collection in future studies. Other limitations of both meta-analyses include heterogeneous populations and variable methodological quality in the included studies, the limited information on potential explanatory factors such as age, and the potential for sex-based diagnostic biases where women with minor symptoms may be less likely to undergo the advanced imaging required to be included in a study of people with confirmed stroke or TIA events.²² Collectively, however, the findings of these meta-analyses suggest that while clinicians should be aware that women may be more likely than men to present with altered level of consciousness or headache and investigate accordingly, current stroke awareness campaigns focusing on traditional symptoms are likely appropriate for both women and men.

Non-standard Abbreviations and acronyms

AHEI

Alternative Health Eating Index

HR

Hazard ratio

IRR

Incidence rate ratio

CI

Confidence interval

OR

Odds ratio

RR

Relative risk

ITT

Intention-to-treat

HA

hysterectomy alone

HBO

hysterectomy plus bilateral oophorectomy

OC

oral contraceptive

HRT

hormone replacement therapy