Article Text
PDF
Abstract
Context The literature provides no clear answer as to whether total oestradiol (E2) concentrations increase the risk of incident cardiovascular disease (CVD) in healthy men.
Objective The authors conducted a systematic review and meta-analysis to estimate the predictive value of E2 for CVD, and to identify study features explaining conflicting results.
Data sources Articles were identified by a Medline and Embase search and citation tracking.
Study selection Eligible articles were prospective population-based cohorts and nested case-control studies on E2 and incident cardiovascular disease (CVD), including myocardial infarction, stroke or death from coronary heart disease.
Data-extraction Independent researchers re-expressed associations of E2 and incident CVD in a uniform manner to be used in meta-regression analyses for identification of study features explaining conflicting results, and to estimate the predictive value of E2 for CVD.
Results and conclusions 14 studies out of 128 electronically identified articles were eligible. Data to be used for meta-analysis could be calculated in seven cases, and in the remaining seven cases, data of three more became available by contacting those authors. Overall, a non-significant association was found with an estimated summary RR of 0.98 for a change of >75th versus <25th percentile in E2 (95% CI 0.74 to 1.31). Mean body mass index (BMI) of the study population (βs −0.8, p<0.004), and quality of E2 assay (βs −0.6, p<0.08) may have modified the relationship between E2 and incident CVD. The present systematic review does not provide evidence for a pronounced harmful or beneficial effect of E2 on risk for incident CVD in healthy men. If present, an effect of E2 on risk for CVD might be modulated by BMI.
- Oestradiol
- testosterone
- cardiovascular diseases
- prospective studies
- meta-analysis
- allied specialties
- endocrinology
Statistics from Altmetric.com
- Oestradiol
- testosterone
- cardiovascular diseases
- prospective studies
- meta-analysis
- allied specialties
- endocrinology
Introduction
Sex steroids are supposed to explain differences in cardiovascular disease (CVD) between men and women. Despite numerous prospective studies during the past 25 years,1–23 the relationship between sex steroids and incident CVD in healthy men is still unclear. Systematic evaluation revealed an increase in risk of low testosterone (T) for CVD in elderly men, but no increase in risk in middle-aged men.24 ,25 Elderly men may have experienced a prolonged negative effect of low T. The relationship between oestradiol (E2) and CVD is still unclear. E2 originates from peripheral conversion of T through enzymatic aromatisation (P450a) mainly in adipose and muscle tissue. The amount/activity of P450a has been suggested to be elevated in males with larger fat mass, and E2 might therefore be increased in obesity.26 In turn, E2 is a major determinant of peripheral T levels. E2 in the systematic circulation reflects the inhibitory tone exerted on gonadotropin release through the hypothalamic-pituitary-gonadal axis.27 E2 may display potentially beneficial as well as harmful effects on the cardiovascular system28 and adipose tissue.29 Clinical treatment of, for example, obesity-related hypogonadism by T substitution therapy affects E2 concentrations as well. At present, the overall risk of joint E2 effects on the cardiovascular system is unknown, that is, it is unrevealed whether circulating concentrations of E2 relate to risk for incident CVD in healthy men.
We thus performed a systematic review according to MOOSE guidelines30; a meta-analysis to estimate the predictive value of endogenous E2 for CVD, and a meta-regression analysis to identify study features that modify the relationship, that is, to explain heterogeneity across studies. Study features hypothesised to be potential modifiers were: study design, E2 assay, time of blood sample, sample size, type of outcome studied, number or type of confounders for which adjustment was made, age of study population, follow-up period, country of study, year of publication and mean body mass index as an increase in adipose tissue that may affect E2 levels.
Methods
Data sources and study selection
Potentially relevant studies were identified through a literature search of EMBASE and MEDLINE (1966–June 2011) using the Medical Subject Headings oestradiol (E2), cardiovascular diseases (CVD), myocardial infarction (MI), stroke (S), atherosclerosis, longitudinal studies and prospective studies. Additional studies were identified by reference lists and citation tracking. Eligible articles were prospective population-based cohorts or nested case-control studies on the relationship between, on the one hand, E2 concentrations (total, free or bioavailable) and on the other hand, CVDs, MI, S or death from coronary heart disease. The search was limited to studies in men who were healthy. Studies on sex steroids in males with a particular disease (heart failure, HIV, etc) were excluded, as interpretation of these studies is complicated because an illness by itself may affect sex steroid concentrations.31
Data extraction
Data were extracted pertaining to E2 concentrations, type of E2 assay involved, blood sample time, sample size, study design (prospective population-based versus nested case-control), type of outcome studied (CVD, cardiovascular mortality (CM), MI, S), number and type of confounders for which adjustment was made, age of study population (mean age), follow-up period, country of study, year of publication and body mass index (BMI) of study population. If investigators presented studies on closely related data, the study with most detailed data, allowing re-expression in a uniform manner (see further), was selected for meta-analysis, that is, references 5 ,7 ,8 ,12 and 17 instead of 18 ,22 ,23 ,19 and 20 or 21, respectively. Articles had to provide enough information to estimate an RR and a 95% CI or an approximation, such as an OR. We extracted the estimated RR, adjusted for the highest number of potentially confounding variables, from each of the original articles. If articles presented insufficient data,3 ,6 ,9 ,14–17 authors were requested by email to provide an adjusted estimate of risk resembling the re-expressed effect size. Quality assessment of individual studies was replaced by exploring sources of heterogeneity by meta-regression analyses. Two authors (GV, PC) independently assessed studies for inclusion and completed data abstraction; disagreements were resolved by consensus with a third author (DDB).
Summary estimate and sources of heterogeneity
We used similar methods as described for a previous meta-analysis.24 ,32 ,33 In short, the original effect size of E2 for CVD of each individual study was re-expressed to a regression coefficient β and SE corresponding to the risk of a change of the upper quartile (>75th percentile) versus the lower quartile (<25th percentile) in total E2 concentrations for the incident CVD outcome at issue, adjusted for the highest number of potentially confounding variables. Analyses were performed only for total E2 (nstudies=10), as the number of studies for free (nstudies=2), or bioavailable (nstudies=1) E2, were too small. Adjusted regression coefficients were combined using a random-effects model. Heterogeneity was investigated with the help of meta-regression analyses, and study features that were consecutively included as possible sources of heterogeneity were the hypothesised items (see Introduction). The quality of the E2 assay of each individual study was estimated by an expert on hormone assays (TF). He evaluated the assay as described in the method section of each individual article blinded for that occasion for results, authors and name of the journal, and rated the assay as poor, low, average or good. Analyses were performed using SPSS V.18.0 software package (SPSS Inc), and Review Manager (RevMan) (computer program), V.5.1.2 Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011.
Results
Eligible articles
Fourteen articles were eligible out of an electronical selection of 128. The desired RR and 95% CIs could be calculated in seven cases,1 ,2 ,4 ,5 ,7 ,8 ,10 and of the remaining seven cases, data of three more articles was obtained by contacting the authors.3 ,6 ,9 Ten articles with data on E2 and incident CVD were, therefore, available for meta-analysis (see figure 1). The studies of total E2, free E2 and bioavailable E2 concentrations described in these articles are listed in table 1.1–10 Only the studies with total E2 concentrations (n=10) were used for the meta-analysis. In the last column, RRs and 95% CIs are presented, after re-expression in a uniform manner, for a change between the upper quartile (>75th percentile) versus the lower quartile (<25th percentile) in total E2 concentrations. Seven articles did not provide sufficient information to be used in meta-analysis.3 ,6 ,9 ,14–17 Authors of these articles were requested by email to provide specific RRs and CIs (per U25/L25) which resulted in the inclusion of another three studies for meta-analysis.3 ,6 ,9 Detailed information of four articles remained inaccessible.14–17
Study flow diagram. 128 electronically identified articles were potentially relevant. After evaluation, 10 articles were directly selected, and 11 additionally by reference lists and citation tracking. Out of the 21 articles, six were excluded as closely related data was reported, three were excluded as progression of atherosclerosis was reported (including them did not essentially make any difference), and another four articles were excluded, as detailed information could not be obtained despite persistent attempts (see Discussion).
Summary of prospective studies of oestradiol and cardiovascular disease in healthy men
Summary estimate and sources of heterogeneity
Overall, no independent effect of E2 on incident CVD was found, as shown by an estimated summary RR (95% CI) of 0.98 (0.74 to 1.31), for a change between the upper quartile (>75th percentile) versus the lower quartile (<25th percentile) in total E2 concentrations (p for heterogeneity <0.0001, Figure 2).Meta-regression analysis of the 10 studies of total E2 concentrations identified ‘mean BMI of study population’, and ‘quality E2 assay’ (borderline significant) as potential sources of heterogeneity (figure 3A,B). The finding may suggest that male populations with a relative low mean BMI (kg/m2) have a favourable risk profile with low E2 concentrations, whereas, populations with a relative increased mean BMI have a favourable risk profile with high E2 concentrations for incident CVD risk. The borderline significant finding of ‘quality E2 assay’ suggests that studies with a better quality of E2 assay show a favourable risk profile with high E2 concentrations.
Meta-analysis of total oestradiol concentrations and incident cardiovascular disease.
Meta-regression analysis of total oestradiol concentrations and incident cardiovascular disease identified ‘mean BMI of study population’ with a standardised β of −0.8 (p<0.004), and ‘quality of oestradiol assay’ with a borderline statistically significant standardised β of −0.6 (p<0.08) as potential sources of heterogeneity (figure 3A and B, respectively).
Confounding and publication bias
The estimated summary RR is based on individual RRs adjusted for the highest number of potentially confounding variables. Most studies adjusted/matched, to some extent, for similar potentially confounding variables: age,1 ,2 ,4–10 obesity,1 ,2 ,4 ,5 ,8–10 smoking,2 ,4 ,5 ,8–10 serum lipids,1 ,4 ,5 ,8 ,9 blood pressure2 ,4 ,5 ,8 ,9 and glucose.4 ,5 ,8 ,9 Neither confounding (specific confounder, total number, classic CVD risk factors), study design (prospective population-based versus nested case–control), type of outcome studied (CVD, CM, MI or S), follow-up period, country of study, time of blood sample, sample size, nor year of publication modified the relationship between E2 and incident CVD. Publication bias was assessed by plotting the number of cases versus effect magnitude: the figure resembled a funnel indicating relative absence of publication bias (figure not shown).34
Discussion
The present systematic review does not provide evidence for a pronounced harmful or beneficial effect of total E2 concentrations (E2) on risk for incident CVD in healthy men. If present, an effect of E2 on risk for CVD might be modulated by BMI.
A hypothesis that might shed light on these findings, and seems worth examining, is a potentially beneficial link between E2 and left ventricular remodelling and function in healthy middle-aged men.35 In contrast, Callou de Sá et al reported an association between increased E2 and coronary artery disease.36 Obviously, cross-sectional studies do not provide information on causality. Indeed, conditions associated with coronary artery disease may as well be associated with increased E2 levels. For example, overweight and obesity are associated with coronary artery disease, which runs parallel to the degree of obesity, with an increase in E2 concentrations.37
E2 may display various effects on cardiovascular and adipose tissue, which might be of particular interest in the presence of overweight.29 On the one hand, E2 may display beneficial effects on endothelial cells, inhibiting migration, proliferation and collagen production in vascular smooth muscle cells, and E2 may improve the function of cardiac myocytes under ischaemic stress.28 On the other hand, E2 could induce inflammatory and pro-coagulant factors, 28 and appear to increase triacylglycerol levels.38 In fact, E2 may play a major role in cell death, although effects on cells of males and females could be different.39
The present study transparently identified sources of heterogeneity (study features explaining conflicting results), namely, BMI and quality of E2 assay (borderline significant). The latter finding underscores the application of more accurate E2 assays, such as mass spectrometry-based methods.40 The former finding might be of clinical importance, although it should be further investigated, as the findings are restricted through limited variation of BMIs of included study populations. Another limitation of the present meta-analysis is that studies were largely performed in healthy white males, with just one study in Asian males,9 and are, therefore, non-informative for non-white males, males with specific diseases or females. Also, some of the articles provided insufficient information, and procedures to estimate RRs (with 95% CI) could not be performed for all selected studies despite a partial successful attempt to gather additional information by correspondence with authors. In any case, findings appeared to be robust and consistent, including three prospective studies on E2 and atherosclerosis into the meta-analysis11–13 which neither changed the estimated summary RR essentially, nor the sources of heterogeneity (data not shown).
In summary, the present systematic review shows an overall neutral effect of E2 on risk of incident CVD in healthy men. Effects on the studied population might be modulated by BMI, or by quality of E2 assay. In future, prospective observational studies, or randomised clinical trials manipulating sex steroid levels, are needed to investigate the different aspects between E2 and CVD, as well as the potential modulating effect of BMI. Mass spectrometry-based methods to assess E2 are additionally valuable, especially if very low E2 concentrations are expected.40
Acknowledgments
We are very grateful to KH Mikulec, CN Lessov-Schlagger, RE Krasnow, E Fransson and M Akishita for their time and willingness to provide additional details on the Twin Study (KM, CL, RK), Gender Study (EF) and Tokyo Study (MA).
References
Footnotes
Funding This work was partly supported by the Flemish Fund for Scientific Research (FWO-Vlaanderen Grant G.0662.07). The study sponsor had no role in study design, analysis or conclusions of the report.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.