Laparoscopically Confirmed Endometriosis and Risk of Incident Stroke: A Prospective Cohort Study

During 2,770,152 person-years of follow-up, there were 893 incident cases of stroke. At baseline in 1989, women with laparoscopically confirmed endometriosis were more likely than women without a history of endometriosis to report a BMI <25 kg/m 2 at age 18, an earlier age at menarche, irregular menstrual cycles, a history of infertility, nulliparity, and to have been past or current oral contraceptive users ( Table 1 ). We observed no difference between women with and without a history of endometriosis for physical activity patterns, alcohol intake, or AHEI dietary intake score. Women with a history of endometriosis were more likely to report a family history of myocardial infarction and stroke. Women with a history of laparoscopically confirmed endometriosis had a 34% greater risk of stroke compared to women without a history of endometriosis in models adjusted for potential confounding factors (HR: 1.34, 95% CI: 1.10–1.62) ( Table 2 ). This association was partially mediated by occurrence of hysterectomy and/or oophorectomy (percent mediated: 39%), postmenopausal hormone therapy (15.5%), age at menopause <45 (12.3%), history of hypertension (8.4%), or history of high cholesterol (4.9%) ( Table 3 ). We observed no difference in the association between endometriosis and risk of stroke by age, BMI, history of infertility, menopausal status, smoking history, history of hysterectomy/oophorectomy, history of hypertension, history of hypercholesterolemia, history of hormone therapy use, and history of age at menopause ( Supplementary Table 1 ). Sensitivity analyses that included all reports of endometriosis, with or without laparoscopic confirmation, did not meaningfully change the results (Data not shown).

The NHSII is an ongoing prospective cohort study that began in 1989 when 116,429 female registered nurses between the ages of 25–42 returned a mailed questionnaire. Participants were recruited from 14 states where the investigators contacted state nursing boards to contact female nurses within the appropriate age range, recruitment details have been described in detail previously 16 . Participants have since moved to all 50 states and have received mailed questionnaires every two years that collected detailed information on a variety of chronic diseases and risk factors. For the present study, NHSII participants were excluded if they had a history of stroke, myocardial infarction, cancer (other than non-melanoma skin cancer) or coronary artery bypass grafting prior to June 1989, when the cohort began or had endometriosis that was not confirmed by laparoscopy. Leaving 112,056 women followed from 1989 until June 2017 ( Supplemental Figure 1 ). The NHSII protocol was approved by the Institutional Review Board (IRB) of the Partners Health Care System, Boston, MA, USA and this analysis was reviewed by the IRB of the University of Arizona. Completion of the baseline and subsequent biennial questionnaires implied consent of the cohort participants. Because of the sensitive nature of the data collected for this study, requests to access the dataset from qualified researchers trained in human subject confidentiality protocols may be sent to Channing Division of Network Medicine ( [email protected] ). This study follows STROBE guidelines 17 . From 1993 and on each biennial questionnaire, participants were asked whether they had physician-diagnosed endometriosis. If participants responded ‘yes,’ they reported the year of diagnosis and whether the endometriosis diagnosis had been confirmed by laparoscopy, the clinical gold standard for endometriosis diagnosis 18 – 20 . Among a subgroup of participants, self-reported endometriosis was validated at two time points in 1994 (n=200) and 2011 (n=711) and a diagnosis of endometriosis was confirmed via medical records of 95%–100% of women reporting laparoscopically confirmed endometriosis in the first and second validations studies, respectively, but in only 56% of women without laparoscopic confirmation 21 . Therefore, we restricted our endometriosis definition to those with laparoscopic confirmation, to reduce misclassification of our exposure. Laparoscopically confirmed endometriosis diagnosis was updated over time, but once a woman reported laparoscopically confirmed endometriosis she was categorized as having a history of endometriosis for the remainder of follow-up. At enrollment in 1989 and biennially, participants reported all incident physician diagnosed “stroke (cerebrovascular accident) or transient ischemic attack” events. Permission was requested from participants or next of kin to obtain and review medical records following self-reported stroke. Stroke was confirmed by physician review of medical records (ICD 430.0–437.0) when available and classified as ischemic or hemorrhagic by the National Survey of Stroke criteria, requiring a typical neurological deficit of rapid or sudden onset lasting ≥24 hours or until death attributable to a vascular cause 22 . For reports of stroke for which medical records were not available/ permitted, strokes were corroborated by nurse participant information. Pathology attributable to infection, trauma, or malignancy was excluded, as were silent strokes discovered only by radiological imaging. On the baseline questionnaire in 1989 and subsequent questionnaires, participants reported a number of health characteristics including their height, current weight, weight at age 18, age at menarche, menstrual cycle pattern in adolescence and currently, parity (number of pregnancies lasting ≥6 months), oral contraceptive (OC) use, smoking history, physical activity (modeled as metabolic equivalents (METs) from recreational and leisure-time activities), race/ethnicity, pre-tax annual household income, family history of myocardial infarction or stroke, history of infertility (>12 months trying to conceive without success), NSAID use, aspirin use, menopausal status, history of oophorectomy or hysterectomy and physician-diagnosed type 2 diabetes, cardiovascular disease, cancer, hypercholesterolemia and/or physician-diagnosed hypertension, and postmenopausal hormone use. All time-varying characteristics were updated every 2 years. Diet in the past year was reported via food frequency questionnaire every 4 years 23 , from which the Alternative Health Eating Index (AHEI) diet score was calculated 24 , 25 . Person-months at risk were calculated from entry into the cohort in 1989 until confirmed i) death ii) cardiovascular event (myocardial infarction, stroke), or iii) their last returned questionnaire, whichever came first. To account for possible confounding by age, all variables presented in Table 1 (aside from age) were standardized to the age distribution of the study sample 26 . Cox proportional hazard models were stratified by calendar time (years) with age (months) as the time metameter and were used to calculate the hazard ratios (HR) and 95% confidence intervals (CI) of incident stroke diagnosis (Model 1). The proportional hazards assumptions were tested using the likelihood-ratio test comparing a model with and without an interaction term for time; they were met. We adjusted for covariates that we hypothesized were potential confounders for the association between endometriosis and risk of stroke 27 with time-varying covariates updated biennially at every questionnaire cycle (Model 2): alcohol intake (0, 0–5, 5+ grams/day), BMI at age 18 (<22, 22.5 to <25, 25 to <30+ kg/m 2 ), current BMI (<22, 22.5 to <25, 25 to <30, 30 to <35, 35+ kg/m 2 ), age at menarche (11 or younger, 12–13, 14+ years of age), menstrual cycle pattern in adolescence (regular, usually irregular, always irregular, no menses), current menstrual cycle pattern in adulthood (regular, irregular, no menses), parity (0, 1, 2, 3+ pregnancies > 6 months), oral contraceptive use history (current, past, never), smoking history (never, past, current), AHEI diet score (quintiles), physical activity (<3, 3 to 8.9, 9 to 17.9, 18 to 26.9, 27+ MET-hour/week), NSAID use (nonuser, current user), aspirin use (nonuser, current user), race (white, non-white), pre-tax annual income (<$50K, $50 to <$100K, $100K+ per year), family history of MI (no, yes), and family history of stroke (no, yes). Mediation analyses were conducted to investigate the proportion of the association between endometriosis and risk of stroke that could be attributed to intermediate variables occurring after endometriosis diagnosis but prior to stroke 28 . The proportion mediated was estimated by comparing the hazard ratio with and without the time-varying proposed mediator 29 , 30 . Potential mediators of interest were: physician-diagnosed hypertension or hypercholesterolemia, age at menopause (≤45 years old, >45 years old), hysterectomy and/or oophorectomy, and hormone therapy use. Effect modification by age and infertility history were observed in prior analyses of endometriosis and cardiometabolic conditions 12 , 31 , therefore we assessed heterogeneity in the association between endometriosis and stroke by potential effect modifiers, including current age (<50 years old, ≥50 years old), BMI (<25, ≥25 kg/m 2 ), history of infertility (yes, no), menopausal status (premenopausal, postmenopausal), smoking status, history of hypertension, history of hypercholesterolemia, history of hormone therapy use, and history of age at menopause (≤45, >45). Women contributed person-time to the appropriate strata given their time-varying status over follow-up. Likelihood ratio tests were used to test for statistically significant differences between groups. 27 Sensitivity analyses were performed to investigate known complexities in endometriosis diagnosis. Women with endometriosis may wait many years between when their symptoms begin and when they receive a diagnosis of endometriosis. Therefore, due to the potential diagnostic delay between endometriosis symptom onset and disease diagnosis 32 , 33 , the date of endometriosis diagnosis was pre-dated in sensitivity analyses by 4, 6, and 8 years. Additionally, we expanded our endometriosis exposure definition to include endometriosis cases both with and without laparoscopic confirmation.

Prior research has suggested that women with endometriosis may be at increased risk of cardiovascular diseases such as myocardial infarction and hypertension. Our prospective analysis with nearly thirty years of longitudinal follow-up suggests that women with laparoscopically confirmed endometriosis may also be at increased risk of stroke. These findings should be replicated in other large, longitudinal cohorts to fully disentangle the contribution of endometriosis treatments on cardiovascular disease risk. Women and their health care providers should be aware of their gynecologic and reproductive history when counseling patients and evaluate cardiovascular risk factors and primary prevention of cardiovascular disease, as well as the signs and symptoms of cardiovascular disease, including stroke.

In this analysis we observed that women with a history of endometriosis had a greater risk for stroke compared to women without a history of endometriosis. This association was found to be partially attributable to the influence of hysterectomy/oophorectomy occurrence, postmenopausal hormone therapy, menopausal status, and history of hypertension. We observed no difference in the association by infertility history, age, or BMI. There are several different pathways through which endometriosis may be associated with risk of stroke. Women with endometriosis have been found to have a hyper-inflammatory milieu both locally (in the peritoneal cavity) and systemically 7 , 9 that may contribute to a greater risk of cardiovascular disease. Specifically, several inflammatory markers have been found to be elevated in women with endometriosis, such as intracellular adhesion molecule 1 (ICAM-1), C-reactive protein (CRP), interleukin-1 and 6 (IL-1 and IL-6), tumor necrosis factor-α (TNF-α) and vascular endothelial growth factor (VEGF) 34 – 36 ; most of these biomarkers have also been associated with cardiovascular disease risk. Additionally, some evidence supports lower risk of cardiovascular disease among those with longer reproductive lifespan 37 and later menopause transition 3 . Early age at menopause, which can be surgically induced by oophorectomy, can substantially modify cardiometabolic disease onset 38 , 39 . Further, there is growing evidence that hysterectomy with or without bilateral oophorectomy may impact cardiovascular disease risk 40 , 41 . Individuals with endometriosis have a higher incidence of hysterectomy with and without oophorectomy and also of surgical menopause, that may contribute to stroke risk 1 , 5 , 6 . We observed that women with endometriosis had a 34% greater risk for stroke compared to women without a history of endometriosis in multivariable models adjusted for detailed potential confounding factors. The findings from this manuscript are in agreement with previous findings. Prior research from NHSII with 20 years of follow-up observed that women with a history of endometriosis had a greater risk of coronary heart disease 12 , hypertension, and hypercholesterolemia 10 . Specifically, women with a history of endometriosis had a 52% greater risk of myocardial infarction, a 91% greater risk of angiographically confirmed angina, a 35% greater risk of coronary artery bypass graft surgery/coronary angioplasty/stent 12 , a 22% greater risk of hypercholesterolemia, and a 29% greater risk of hypertension 10 . Research from a retrospective cohort in Taiwan (median follow-up: 9.2 years), observed that women with endometriosis had a 1.2-fold greater risk of any major adverse cardiovascular and cerebrovascular events compared to women without endometriosis, with a 1.16-fold greater risk of cerebrovascular accident 13 . Similarly, a retrospective cohort from the Health Improvement Network (THIN) in the United Kingdom (average follow-up: 5.7 years, maximum follow-up: 23 years) observed that women with endometriosis had a 1.24-fold greater risk of their composite CVD endpoint, and a 1.19-fold greater risk of cerebrovascular disease. The cross-sectional Japan Nurses’ Health Study reported a two-fold risk of either transient ischemic attack or cerebral infarction for women with endometriosis compared to women without endometriosis 11 . We observed that the association between endometriosis and risk of stroke was partially mediated by occurrence of oophorectomy and/or hysterectomy, hormone therapy, age at menopause, and history of hypertension. Prior research from the UK observed that when women with hysterectomy and oophorectomy were excluded from analyses, the relationship between endometriosis and CVD was attenuated, 14 implying that gynecologic surgery, or the indication for the surgery, may partially contributed to increased risk of cardiovascular disease. Mu’s analysis in the NHSII observed that 42% of the association between endometriosis and coronary heart disease was mediated by hysterectomy/oophorectomy and age at surgery 12 . Prior research has suggested that hysterectomy with and without ovarian conservation may influence cardiovascular disease risk 38 , 40 – 42 . Research has also observed that women with endometriosis may have greater risk of hypertension than women without endometriosis 10 , 11 . In our analysis, hypertension was found to be a modest mediator of the association between endometriosis and risk of stroke. While prior research in the NHSII observed the association between endometriosis and coronary heart disease was strongest among younger women (<40 years of age) 12 , we observed no difference in the relationship between endometriosis and risk of stroke by age (<50 years of age). This study has many strengths including its prospective design, large sample size, endometriosis confirmed by laparoscopy, nearly thirty years of longitudinal follow-up, and detailed evaluation of potential time-varying confounding, mediation, and effect modification of the association between endometriosis and stroke. However, there are also important limitations that must be considered. The exact onset of endometriosis cannot be determined – regardless of study design or population sampled, and the time from symptom onset to definitive diagnosis (laparoscopy) can be many years (NHSII mean=4 years, general population mean= 7 years 32 ). To reduce misclassification of endometriosis, we restricted our analyses to laparoscopic confirmation of endometriosis which has extremely high validity with the medical record (≥96%) 21 . We also conducted two sensitivity analyses related to our endometriosis definition – we predated our endometriosis diagnoses by 4, 6, and 8 years, and we also expanded our endometriosis definition to include all women with any self-reported endometriosis. In both sensitivity analyses, the results did not meaningfully change. We were unable to differentiate between hemorrhagic and ischemic stroke using the current data set nor was ischemic stroke subtype available for analysis. Whether or not endometriosis is more strongly associated with a particular stroke subtype is an important area for future research that could inform strategies to reduce stroke risk among patients with endometriosis. Moreover, given limited statistical power in this data source, we are unable to investigate the influence of serious cardiovascular events, such as myocardial infarction and coronary artery bypass graft surgery, on the association between endometriosis and stroke risk. Our unexposed group may include women who have never reported, recognized, or received a diagnosis for their endometriosis. We expect the prevalence of undiagnosed endometriosis to be sufficiently low and to have minimal impact on study results, however, their characteristics will be diluted among the ~80,000 true unexposed women 43 . Our analysis was able to take into account the influence of hysterectomy and oophorectomy on risk of stroke, however we were not able to incorporate information on all treatments for endometriosis that may influence stroke risk. The NHSII cohort is not a random sample of US women; thus, our findings may not be generalizable to the entire population. However, it is unlikely that the biologic associations observed in this cohort will differ from women in general 16 , 44 . The high level of education and interest in health are distinct advantages that aid our ability to collect valid, high-quality information and reduce possible confounding by socioeconomic factors. Our findings, along with previous research on endometriosis and CVD, suggest that clinicians should consider both reproductive and gynecologic health history when counseling patients regarding their CVD risk. Future research should focus on replicating these findings, looking separately at ischemic vs. hemorrhagic stroke, and assessing the influence of reducing cardiovascular disease risk factors among women with endometriosis.

Endometriosis is a chronic inflammatory gynecologic condition that burdens approximately 10% of women. 1 Cardiovascular disease is known to present differently among men and women, including later age at onset for women 2 , 3 , and a higher incidence of stroke as the first cardiovascular event in women 4 . Female specific risk factors for cardiovascular disease are increasingly being recognized. Mounting evidence suggests that women with endometriosis may be at greater risk for cardiovascular diseases later in life 5 , 6 . Endometriosis may influence risk of cardiometabolic diseases, including stroke, through alterations in the endogenous inflammatory, immunologic, and hormonal milieu 7 – 9 . Further, treatments for endometriosis, such as hormonal medications, hysterectomy, and oophorectomy, may modify cardiovascular disease risk 6 . Previous research has suggested that women with endometriosis may have greater risk of hypertension 10 , 11 , hypercholesterolemia 10 , and coronary heart disease 12 , 13 . Despite plausible mechanisms and prior research on coronary heart disease that suggest a possible association between endometriosis and elevated risk of stroke, there is a paucity of prospective research with longitudinal follow-up for stroke. Recently, analyses from the Health Improvement Network database in the UK observed that women with endometriosis had a 19% increased risk of cerebrovascular disease 14 . Additionally, retrospective data from Taiwan National Health Insurance reported approximately a 16% greater risk of acute ischemic or hemorrhagic stroke for women with endometriosis compared to women without endometriosis 13 . However, current research on endometriosis and risk of long-term health outcomes, including stroke, is limited by short durations of follow-up, limited accounting for potential confounding factors and reproductive health history, and/or cross-sectional analyses ignoring temporality that negate potential for causal inference 6 , 15 . To overcome these prior limitations, the current study investigated the association between endometriosis and risk of incident stroke among participants in the Nurses’ Health study II (NHSII) who have been followed for nearly 30 years.