Introduction
The term transgender is ascribed to individuals who transcend culturally defined categories of gender. Their gender identity, expression, or behavior are not congruent with their natal sex. Reference Coleman, Bockting, Botzer, Decuypere and Feldman1,Reference Lahue, Torres, Rosendale and Singh2 This is in contrast to cisgenders whose gender identity aligns with their sex assigned at birth. Reference Cicero, Reisner, Merwin, Humphreys and Silva3 Male-to-female (MTF) transgenders or transwomen are those whose sex was assigned male, but who identify as female. Reference Coleman, Bockting, Botzer, Decuypere and Feldman1 Recent estimates for MTF transgenders are at 6.8 in 100,000 individuals. Reference Arcelus, Bouman, Van Den Noortgate, Claes, Witcomb and Fernandez-Aranda4
Despite increasing gains in rights, there remain enormous gaps in addressing the health needs of transgender individuals. Reference Lahue, Torres, Rosendale and Singh2,Reference Streed, Harfouch, Marvel, Blumenthal, Martin and Mukherjee5 They suffer from stigma and high rates of depression, anxiety, and HIV infection. Reference Safer, Coleman and Feldman6 Though there have been more studies on the transgender population in recent years, the majority of health research for transgenders has focused on mental health, cross-sex hormone therapy, health-harming behaviors, and HIV/AIDS. Data on chronic health conditions including cerebrovascular diseases is scant. Reference Cicero, Reisner, Merwin, Humphreys and Silva3 There is also a dearth of data and guidelines on the appropriate medical care for transgender patients. Reference Lahue, Torres, Rosendale and Singh2
Though multiple studies have established the adverse outcomes of hormonal therapy in ciswomen, crucial differences in clinical characteristics exist between ciswomen and transwomen. It is, therefore, not apt to extrapolate these findings to transgender individuals. Reference Lahue, Torres, Rosendale and Singh2 Research into the epidemiology and characterization of cerebrovascular diseases in this special population is warranted.
Our study aims to systematically review and report on existing data on stroke in MTF transgenders. Where possible, we hope to describe the clinical characteristics, hormone regimen exposures, and outcomes of MTF transgenders who suffered from stroke.
Methodology
Standard Protocol Approvals, Registrations, and Patient Consents
We performed this meta-analysis in concurrence with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) and Meta-Analyses of Observational Studies in Epidemiology (MOOSE). Reference Moher, Liberati, Tetzlaff and Altman7,Reference Stroup8 We registered the protocol of this review in PROSPERO (CRD42021224819).
Criteria for Selection of Studies
We considered cohort studies, case–control studies, case series, and case reports for this review. We included studies that reported information on the occurrence of stroke in transgender MTF. Exposure included hormonal therapy, i.e. estrogen with or without antiandrogens regardless of dosage. The development of stroke in MTF transgenders was considered to be the outcome of interest. There were no restrictions implemented in terms of age, sex, and ethnicity of the population. We included studies written in the English language.
Search Methods for the Identification of Studies and Selection of Studies
We performed a comprehensive systematic search of records in major scientific databases including MEDLINE by PubMed, CENTRAL by Cochrane, Scopus, Embase, ClinicalTrials.gov, Web of Science, and HERDIN from inception until November 2020. Search strategies (detailed in Appendix) were developed using Medical Subject Headings (MeSH) and text words related to transgender persons and sex steroids or hormone therapy. The outcome of interest included terms pertaining to stroke or cerebrovascular disease. To ensure literature saturation, hand searching of additional studies was done by going through the reference section of included studies and relevant reviews. Personal contacts to authors who published articles were made to obtain further data.
Two investigators (KHDI and JDBD) searched independently to identify relevant articles using the search term strategies developed. After deduplication, the titles and abstracts of the remaining studies were assessed by two authors using predetermined screening criteria. Full-text articles of studies that fulfilled the screening criteria were retrieved and evaluated using predetermined eligibility criteria. Disagreements were solved with the contribution of a third reviewer (AIE) and via consensus. Finally, studies that satisfied the eligibility criteria were included in the quantitative analyses.
Methodological Quality Assessment of the Included Studies
We used the Newcastle–Ottawa Scale (NOS) to evaluate the quality of observational studies included in the review (see Supplementary Material). Reference Wells, Shea and O’Connell9 The risk of bias was independently assessed by two investigators (KHDI and JDBD) while disagreements were resolved with the contribution of two other reviewers (AIE and MCPF) and by consensus. For the present review, factors that would make a study have a low risk for bias included clearly defined selection criteria, representativeness of the sample of transgender patients, and appropriate ascertainment of outcomes.
We used the Murad tool to evaluate the risk of bias in noncomparative cohorts and case series/reports. We considered “poor,” “moderate,” or “good” quality when 3 or fewer, 4, or 5 of the criteria were fulfilled, respectively. Reference Murad, Sultan, Haffar and Bazerbachi10 Three investigators (KHDI, AIE, and JDBD) evaluated the methodological quality of the included studies; discrepancies were resolved by consensus and discussion with a fourth reviewer (MCPF).
Data Collection
Data from the included studies were extracted by two investigators (KHDI and JDBD) using standardized forms, which included the following information regarding the study: title, citation, setting, design and duration, the total population of patients in the study, the number of patients in the population of interest, comorbidities, hormonal regimens taken, clinical characteristics, and outcomes.
Statistical Analyses
To summarize data, we used frequencies and proportions for categorical variables while means (standard deviation) or medians (range) were used for continuous variables. We expressed the data in 95% confidence intervals. We considered the individual patient as the unit of analysis. We performed the meta-analysis of proportions by utilizing the following R (Version 3.6.3) packages: metafor, meta, and weightr. We included studies with > 5 included patients in the meta-analysis. The proportions were pooled by the random-effects model measured using the restricted maximum likelihood estimator. The data for computed proportions that fell in the range of < 0.20 or > 0.80 were converted using logit transformation. This was done to improve their statistical properties. We performed leave-one-out sensitivity analyses as well as a set of case deletion diagnostics based on linear regression analyses to assess the robustness of summary estimates on the effect of outliers. In addition, we constructed a Baujat plot to mark studies largely influencing the summary statistics. To evaluate the heterogeneity of pooled estimates, we employed the chi-squared test (χ2; p-value < 0.10 to detect significant heterogeneity) and I2 tests with >25%, >50%, and >75% considered as low, moderate, or high degree of heterogeneity, respectively.
Results
Included Studies
We identified a total of 492 studies from the electronic database search. No relevant studies were identified from Clinicaltrials.gov and from HERDIN. A total of 373 articles remained after excluding duplicates. After assessing the titles and abstracts, we excluded 358 due to non-relevance. The full texts of 18 articles were subjected to eligibility criteria. Fourteen studies were included in the qualitative analyses while 5 of these studies were included in the quantitative analyses (Figure 1). The studies by Asscheman et al. 1989 and 2011 were not included in the quantitative analysis due to an overlap with the population of van Kesteren et al. 1997 (same but expanded cohort). Reference Asscheman, Gooren and Eklund11-Reference Asscheman, Giltay, Megens, de Ronde, van Trotsenburg and Gooren13 Asscheman et al. 2011 was also excluded since they reported on stroke as a cause of mortality instead of morbidity Reference Asscheman, Giltay, Megens, de Ronde, van Trotsenburg and Gooren13 . Nokoff et al. 2018 was excluded from the quantitative analysis since it was a community-based study where data were taken from self-reported surveys. Reference Nokoff, Scarbro, Juarez-Colunga, Moreau and Kempe14
PRISMA flow diagram for study selection.
Assessment of Risk of Bias
We appraised three comparative cohort studies using the NOS, which revealed Wierckx et. al., 2013 and Getahun 2018 et. al., 2018 to be of good quality while the study of Nokoff et al. 2019 was of poor quality. Evaluation of the remaining observational studies was performed using the Murad tool. Reference Murad, Sultan, Haffar and Bazerbachi10 Five studies were assessed to be of good quality while the rest were of moderate quality (see Supplementary Material.)
Population, Exposure and Outcome Characteristics in the Included Studies
A total of 14 studies reported MTF transgenders who developed stroke. There were 8 retrospective cohort studies, 5 case reports, 1 cross-sectional survey, and 1 case series (Table 1). Inclusive years of cohort studies ranged from 8 to 43 years (Mean 26 years).
Table of included studies
* Included in quantitative analysis.
Table 2 presents data on the MTF cohorts as a whole. Details regarding the age, hormonal therapy regimens, and baseline cardiovascular risk factors indicated therein refer to the entire MTF cohort of each study. Reference Asscheman, Gooren and Eklund11-Reference James, Chang and Imhof18 The sizes of the MTF populations ranged from 49 to 2842 (Mean 1002) while the number of patients who suffered a cerebrovascular event ranged from 1 to 53 (Total 109, Mean 14). Only three studies reported types of stroke. Reference Asscheman, Gooren and Eklund11,Reference Van Kesteren, Asscheman, Megens and Gooren12,Reference Getahun, Nash and Flanders16 Only a handful of studies were able to specify details regarding hormonal therapy and cardiovascular risk factors in these patients. MTF transgenders received varied preparations of estrogen either orally, transdermally, or via injection with or without antiandrogens. Only two cohort studies reported the duration of hormone therapy in the MTF cohort: 19.4 years ± 7.7 in Asscheman et. al., 2011 and 6.0 years (3–11 years) in Wierckx et. al., 2013. Getahun et al. 2018 was the sole study that reported the average maximum daily dosage of estradiol in MTF transgenders who suffered from either VTE or ischemic stroke (4.1 mg; range 1–10mg).
Population characteristics of MTF transgenders identified in cohort studies
ICH = Intracerebral hemorrhage; IM = Intramuscular; GnRH = Gonadotropin-releasing hormone; MTF = Male-to-female; NR = Not reported; PO = Per orem; TIA = Transient ischemic attack.
* Data refers to the entire transgender cohort.
Table 3 summarizes data from case series, case reports, and cohort studies that reported details regarding MTF individuals who suffered from stroke. Reference Lahue, Torres, Rosendale and Singh2,Reference Asscheman, Giltay, Megens, de Ronde, van Trotsenburg and Gooren13,Reference Wierckx, Elaut and Declercq15,Reference deMarinis19-Reference Kwan and Pikula23 Of the 23 patients reported, only 13 patients had data regarding stroke type. Reference Lahue, Torres, Rosendale and Singh2,Reference deMarinis19–Reference Kwan and Pikula23 The majority of the strokes were ischemic in nature (10 out of 14) with varying locations. One patient suffered from concurrent ischemic stroke and subarachnoid hemorrhage. Reference Lahue, Torres, Rosendale and Singh2 Hormone use in these individuals was relatively high and ranged from 2 to 32 years. The patients were notably young with 19 individuals being under 60 years of age. Risk factors such as prior stroke, hypertension, diabetes, and smoking were identified in 16 patients, whereas 2 case reports reported no known risk factors in their patients. Reference deMarinis19,Reference Biller and Saver20
Clinical characteristics of MTF transgenders who suffered from stroke
DM = diabetes mellitus; HTN = Hypertension; ICH = Intracerebral hemorrhage; NR = Not reported; SAH = Subarachnoid hemorrhage; T2DM = Type 2 diabetes mellitus; TIA = Transient ischemic attack.
Frequency of Cerebrovascular Events in MTF Transgenders
Institution-based cohort studies that reported the development of cerebrovascular events in MTF transgenders’ ongoing hormonal therapy were included in the quantitative analysis. The term cerebrovascular events refer to any stroke type, i.e. transient ischemic attack, ischemic stroke, hemorrhagic stroke, or subarachnoid hemorrhage. Random-effect modeling analysis for the frequency of cerebrovascular events in MTF transgenders showed an overall estimate of 2% (Figure 2). The analysis detected a moderate degree of heterogeneity (I2 = 62%). This could be due to the heterogeneity of stroke types included in the meta-analysis. Subgroup analysis was not done since only five studies were included.
Random-effect modeling analysis for frequency of cerebrovascular events in MTF transgenders.
Discussion
Random-effect modeling revealed an overall frequency of 2% for cerebrovascular events among MTF transgenders. However, epidemiological data on the incidence and clinical characteristics of cerebrovascular disorders in MTF transgenders is still lacking. Details, albeit limited, regarding the clinical characteristics of these patients were also reported in the present study. The majority were under 60 years of age. Hormonal therapy regimens varied and included estrogens with or without antiandrogens taken for a duration of 2–32 years. Ischemic stroke was the most commonly reported cerebrovascular event.
The Global Burden of Disease Lifetime Risk of Stroke Collaborators in 2016 reported a 24.5% global lifetime risk of stroke in 2016. The risk among men was 24.7% almost similar to the risk among women at 25.1%. Ischemic stroke risk was higher compared to hemorrhagic stroke risk. 24 Meanwhile, the risk of stroke for people aged 18–50 has been estimated at 10%–15%. Reference Ekker, Verhoeven, Vaartjes, van Nieuwenhuizen, Klijn and de Leeuw25 Similar to global data, the present study revealed more reports of ischemic stroke compared to hemorrhagic stroke. However, the estimated frequency arrived at in the present study is much smaller compared to the aforementioned lifetime stroke risk worldwide (2% vs. 24.5%) as well as compared to the risk of stroke in the young (2% vs. 10%–15%). Issues that relate to the precise definition of transgender status, the predominance of hospital-based data, which may bias toward those individuals with better health-seeking behavior, and the gap in healthcare access of transgender individuals may account for this lower prevalence of cerebrovascular disease in the MTF population.
Formal epidemiologic studies on the incidence and prevalence of transgenderism are faced with methodological issues such as in ascertainment of transgender status. Reference Coleman, Bockting and Botzer26 A higher prevalence is yielded with self-identification as opposed to utilizing precise diagnostic criteria for gender dysphoria. Reference Zucker27 This is reflected in the present study in which some studies used the criteria for gender dysphoria, Reference Asscheman, Gooren and Eklund11,Reference Van Kesteren, Asscheman, Megens and Gooren12 others used DSM-III-3, DSM-IV, and ICD codes Reference Wierckx, Elaut and Declercq15,Reference Getahun, Nash and Flanders16 while other studies defined transgender status as individuals referred to specialist or gender clinics. Reference Asscheman, Giltay, Megens, de Ronde, van Trotsenburg and Gooren13,Reference Nota, Wiepjes, de Blok, Gooren, Kreukels and den Heijer17,Reference James, Chang and Imhof18 These methodological issues could have affected the precision of the estimates of frequency in the present study.
Recent studies also suggest that the prevalence rates of self-reported transgender identity are higher compared to prevalence rates based on samples of clinic-referred adults. Reference Zucker27 In the present study, the reported populations were all clinic or hospital-based save for one study that used a self-reported survey. Reference Nokoff, Scarbro, Juarez-Colunga, Moreau and Kempe14 Study settings were mostly in European countries or the USA. In addition to the gap in medical knowledge for transgender care, financial, health systems, and socioeconomic barriers exist. It is evident that more population-based studies across the globe that rely on samples taken outside the clinic or hospital setting are needed Reference Zucker27 . These studies could help close the health equity gap affecting transgenders more so if they proposed potential solutions to address identified healthcare barriers. Reference Safer, Coleman and Feldman6
In men, estrogen has a questionable role. Some studies suggest that there may be a particular physiologic threshold of circulating androgen levels that protect against ischemic brain injury. Reference Boese, Kim, Yin, Lee and Hamblin28 In women, estrogen administration and the effects of hormonal therapy are more established. Reference Van Kesteren, Asscheman, Megens and Gooren12 Low estrogen levels increase the risk of cardiovascular disease and accelerate atherothrombosis in women below 40 years of age. Hormonal therapy meanwhile in postmenopausal women may increase the risk of cardiovascular disease. Reference Morselli, Santos, Criollo, Nelson, Palmer and Clegg29 The same may be true for cross-sex hormone therapy in transwomen. Reference Wierckx, Elaut and Declercq15
Recent long-term follow-up studies have shown that estrogen and antiandrogen therapy may negatively impact cardiovascular health in transwomen. Reference Wierckx, Elaut and Declercq15 A higher cardiovascular mortality rate has been reported in transwomen as opposed to the general population. Reference Van Kesteren, Asscheman, Megens and Gooren12 A handful of studies have also suggested that estrogen use in MTF transgenders confers increased risks of venous thromboembolism and pulmonary embolism Reference Wierckx, Elaut and Declercq15 . However, the mechanisms by which sex steroids may alter vascular physiology and hormone receptor status in transgenders remain to be elucidated.
Two cohort studies identified in the present study compared transgender populations to control cismales and cisfemales. MTF transgenders were reported to have a higher prevalence of cerebrovascular diseases compared to their cismale and cisfemale counterparts. Reference Getahun, Nash and Flanders16 Getahun et. al.’s findings also suggested that prolonged hormone intake increased the risk for ischemic stroke events with the risk becoming substantially higher in those taking estrogen after 6 years or more of follow-up (10-fold higher risk vs. cisgender males and 4-fold higher risk vs. cisgender females). Reference Getahun, Nash and Flanders16
Limitations of the present review include the paucity of studies available on cerebrovascular diseases in transgenders, the small sample sizes of the included studies, and convenience-based sampling methods used in the studies. An association between hormonal therapy use and stroke cannot be made based on this study. Furthermore, the relationship between cross-sex hormone treatment and cardiovascular risk profile in MTF transgenders is complex. The varied preparations, routes of administration, and dosages of hormonal therapy, which were not given in a standardized manner may all have an effect. Reference Wierckx, Elaut and Declercq15
It is clear that further research is needed to determine how hormonal therapy may influence stroke risk within the transgender population, especially among MTF transgenders. Hormonal therapy for cis-gendered populations is not a guide for cross-sex hormone treatment for transgenders who have a unique physiology and distinct risk factors.
Conclusion
There are limited studies in the existing literature that have reported on cerebrovascular diseases in the transgender population. Even fewer studies have described the clinical characteristics and outcomes of stroke in this particular population subset. Though we estimated the frequency of cerebrovascular events in MTF transgenders in our quantitative analysis, our results may have been subject to selection bias, insufficient exposure ascertainment, and issues on the true causality of stroke in these patients. Healthcare issues of transgender individuals have slowly come to light in recent years, but more studies on the epidemiology and clinical profile of cerebrovascular diseases in this special population are warranted. These studies will aid us in coming up with better treatment guidelines for preventive health care and management of cerebrovascular diseases to address the specific needs of MTF transgenders.
Disclosures
The authors have nothing to disclose.
Statement of Authorship
KHDI: Conceptualization, data curation, formal analysis, interpretation of data, writing – original draft, writing – review and editing.
JDBD: Conceptualization, data curation, formal analysis, interpretation of data, writing – original draft, writing – review and editing.
AIE: Conceptualization, data curation, formal analysis, interpretation of data, writing – original draft, writing – review and editing.
MCPF: Conceptualization, data curation, formal analysis, interpretation of data, writing – original draft, writing – review and editing.
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/cjn.2021.54.
