Schuit SCE, Oei HS, Witteman JCM, Geurts van Kessel CH, van Meurs JBJ, Nijhuis RL, van Leeuwen JPTM, de Jong FH, Zillikens MC, Hofman A, Pols HAP, Uitterlinden AG. Estrogen Receptor α Gene Polymorphisms and Risk of Myocardial Infarction. JAMA. 2004;291(24):2969-2977. doi:10.1001/jama.291.24.2969
Author Affiliations: Departments of Internal Medicine (Drs Schuit, van Meurs, van Leeuwen, de Jong, Zillikens, Pols, and Uitterlinden and Ms Geurts van Kessel) and Epidemiology & Biostatistics (Drs Schuit, Oei, Witteman, Nijhuis, Hofman, and Pols), Erasmus Medical Center, Rotterdam, the Netherlands.
Context The role of estrogens in ischemic heart disease (IHD) is uncertain.
Evidence suggests that genetic variations in the estrogen receptor α (ESR1) gene may influence IHD risk, but the role of common
sequence variations in the ESR1 gene is unclear.
Objective To determine whether the ESR1 haplotype created
by the c.454-397T>C (PvuII)
and c.454-351A>G (XbaI)
polymorphisms is associated with myocardial infarction (MI) and IHD risk.
Design, Setting, and Participants In 2617 men and 3791 postmenopausal women from The Rotterdam Study (enrollment
between 1989-1993 and follow-up to January 2000), a population-based, prospective
cohort study of participants aged 55 years and older, ESR1
c.454-397T>C and c.454-351A>G haplotypes were
determined. Detailed interviews and physical examinations were performed,
blood samples were obtained, and cardiovascular risk factors were assessed.
Main Outcome Measure The primary outcome was MI and IHD defined as MIs, revascularization
procedures, and IHD mortality.
Results Approximately 29% of women and 28.2% of men were homozygous carriers
of the ESR1 haplotype 1 (−397 T and −351 A) allele, 49% of women and
50% of men were heterozygous carriers, and 22% of women and 21.4% of men were
noncarriers. During a mean follow-up of 7.0 years, 285 participants (115 women;
170 men) had MI, and 440 (168 women; 272 men) had an IHD event, of which 97
were fatal. After adjustment for known cardiovascular risk factors, female
heterozygous carriers of haplotype 1 had an increased risk of MI (event rate,
2.8%; relative risk [RR], 2.23; 95% confidence interval [CI], 1.13-4.43) compared
with noncarriers (event rate, 1.3%), whereas homozygous carriers had an increased
risk (event rate, 3.2%; RR, 2.48; 95% CI, 1.22-5.03). For IHD events, we observed
a similar association. In women, the effect of haplotype 1 on fatal IHD was
larger than on nonfatal IHD. In men, the ESR1 haplotypes
were not associated with an increased risk of MI (event rate, 5.7%; RR, 0.93;
95% CI, 0.59-1.46 for heterozygous carriers; and event rate, 5.1%; RR, 0.82;
95% CI, 0.49-1.38 for homozygous carriers) compared with noncarriers (event
rate, 5.8%) and were not associated with an increased risk of IHD.
Conclusions In this population-based, prospective cohort study, postmenopausal women
who carry ESR1 haplotype 1 (c.454-397
T allele and c.454-351 A allele) have an increased
risk of MI and IHD, independent of known cardiovascular risk factors. In men,
no association was observed.
Ischemic heart disease (IHD) has a strong genetic component, but the
identity of the genetic risk factors is unknown. There is a large ongoing
effort to find genes involved in cardiovascular disease. In a large case-control
study that examined 112 polymorphisms in 71 candidate genes for myocardial
infarction, 3 associated gene variants were identified.1 However,
genes involved in the pathways of sex steroids were not considered.
Several lines of evidence implicate sex hormones in cardiovascular disease
risk, such as the difference in disease risk between men and women. The risk
of IHD in women between puberty and menopause is lower than that in age-matched
men. However, this sex difference diminishes when postmenopausal women and
men of similar age are compared.2,3 These
observations have led to the suggestion that decreasing endogenous estrogen
after menopause may be the critical factor in removing the relative protection
against IHD that women have in their premenopausal years.
Estrogen exerts its effects by binding to the estrogen receptors α
and β that, once activated, regulate the expression of multiple genes.
A large body of data implicates the estrogen receptor α gene (ESR1) in cardiovascular disease. In 1997, Sudhir et al described a
man with a null mutation in the ESR1, leading to
unresponsiveness to estrogen. This 31-year-old man was reported to have premature
atherosclerotic coronary artery disease and endothelial dysfunction despite
the presence of high levels of circulating estrogen.4 Furthermore, ESR1 has been identified in most cardiovascular tissues
such as the coronary arterial wall in smooth muscle cells,5,6 endothelial
cells,7 and myocardial cells.8 In
addition, fewer estrogen receptors were found in premenopausal women with
atherosclerotic coronary arteries than in those with normal coronary arteries.5 Finally, variant ESR1 transcripts
are extensively expressed in human vascular tissues.9
It is conceivable that common sequence variations (polymorphisms) in
the ESR1 gene affect cardiovascular disease risk
in the general population. Several single-nucleotide polymorphisms (SNPs)
and variable-number tandem repeat polymorphisms have been identified in the ESR1 gene (http://www.ncbi.nlm.nih.gov). Cross-sectional
studies have reported associations between a number of these polymorphisms
in the ESR1 gene and cardiovascular risk factors
and phenotypes, including body mass index (BMI),10 hypertension,11 coronary flow reserve,12 and
Of the polymorphisms identified in the ESR1 gene,
the c.454-397T>C (also known as IVS1-397 T/C, rs2234693,
and the PvuII restriction site) and c.454-351A>G SNPs (also known as IVS1-351 A/G, rs9340799, and the XbaI restriction site) are the most widely studied so far.
These polymorphisms are located in the first intron of the ESR1 gene, 397 and 351 base pairs upstream of exon 2. Herrington et
al15 have recently shown a potential functional
significance for the c.454-397T>C polymorphism. The
aim of this study was to determine whether these polymorphisms in the ESR1 gene are associated with incident IHD and myocardial
The Rotterdam Study is a population-based prospective cohort study of
men and women that was initiated to assess prevalence, incidence, and determinants
of diseases in the elderly.16 The main focus
of the Rotterdam Study was on cardiovascular, neurogeriatric, ophthalmologic,
and locomotor diseases. Between July 1, 1989, and May 17, 1993, all individuals
aged 55 years or older and who were residents of Ommoord, a district of Rotterdam,
the Netherlands, were invited to participate. A total of 7983 men and women
(78% of those eligible) entered the study, and 7085 participants visited the
research center. Baseline examinations took place between July 5, 1989, and
September 21, 1993, and included an initial home visit and interview by a
trained research assistant and an extensive physical examination at the research
center. The Rotterdam Study was approved by the medical ethics committee of
the Erasmus Medical Center. Each eligible person received written and oral
information about the goals and research methods of the study, together with
a description of the examinations involved. Written informed consent was obtained
from all participants.
Our study population of 6408 participants (3791 women) included white
men and postmenopausal white women who were able to visit the research center
and who completed all parts of the baseline examination, including a blood
Cardiovascular risk factors were obtained by interview and physical
examination at baseline. Interview information, including smoking habits,
age at menopause, and use of hormone therapy, was obtained by a trained research
assistant. Hormone therapy was defined as current or former user and never
user. Smoking was categorized as current, past, or never smoker.
Anthropometric measurements were obtained at the research center. Body
mass index was calculated as weight in kilograms divided by height in meters
squared. Two standardized blood pressure measurements were taken by using
a random zero sphygmomanometer, with the participant in sitting position,
and averaged. Hypertension was defined as a systolic pressure of at least
160 mm Hg or a diastolic pressure of at least 100 mm Hg or use of antihypertensive
medication, encompassing grades 2 and 3 hypertension, according to the World
Health Organization (WHO) criteria.17 After
an overnight fast, blood samples were obtained. Serum total cholesterol level
was determined by an enzymatic procedure.18 High-density
lipoprotein (HDL) level was measured similarly, after precipitation of the
non-HDL fraction. Diabetes mellitus was considered present with current use
of antidiabetic medication or a nonfasting or postload glucose level above
198 mg/dL (11 mmol/L), according to the WHO.19
A history of myocardial infarction was based on self-reported information
verified with the general practitioner's or hospital records, which included
written information on diagnosis and treatment or electrocardiographic (ECG)
evidence. Infarctions detected by the Modular ECG Analysis System without
evidence of symptoms (silent myocardial infarctions) were verified by an experienced
Follow-up started at inclusion into the study and ended either on January
1, 2000, or at the participant's death, whichever was earlier. Research assistants
collected follow-up data on cardiovascular disease morbidity and mortality
from the general practitioners, and in case of treatment by a specialist,
hospital records were retrieved. Information on vital status of the participants
was obtained regularly from the municipal health authorities in Rotterdam.
All collected events were verified by review of hospital discharge reports
and letters from medical specialists. Two research physicians independently
coded events according to the International Classification
of Diseases, 10th Revision (ICD-10).22 In
case of discrepancy, consensus was attained in a separate session. A medical
expert in cardiovascular disease also reviewed all coded events for final
classification. In the analyses, we used the following outcome measurements:
myocardial infarction (I21) and IHD (defined as myocardial infarction [I21],
percutaneous transluminal coronary angioplasty [PTCA; Z95.5], coronary artery
bypass graft surgery [CABG; Z95.1], and death from IHD [I20-I25]). In identifying
myocardial infarctions, general practitioner and hospital records were reviewed,
and all available information, which included ECG, cardiac enzyme levels,
and the clinical judgment of the treating specialist, was used to code the
events. Silent myocardial infarctions were not included in the analysis. Revascularization
procedures were identified by review of hospital discharge letters from the
medical specialist. For further analyses, myocardial infarction and IHD were
also classified as fatal and nonfatal. In addition, all-cause mortality was
also documented during follow-up.
All participants were genotyped for the c.454-397T>C and c.454-351A>G polymorphisms. We described
the polymorphisms in relation to a specific human ESR1 complementary
DNA sequence (accession number NM_000125), in which position 454 of the protein
coding sequence is the first nucleotide of the start of the next closest exon
to the polymorphisms studied (exon 2). The variations were 397 and 351 nucleotides
upstream in the intron. These polymorphisms have also been described at http://www.ncbi.nlm.nih.gov/SNP under identification numbers rs2234693
(c.454-397T>C) and rs9340799 (c.454-351A>G).
DNA was extracted with proteinase K and sodium dodecyl sulfate digestion
at 37°C overnight and purified with phenol-chloroform extractions. The
extracted DNA was then precipitated with NaCl at 4 mol/L and 2 volumes of
cold absolute ethanol. DNA was solubilized in double-distilled water and stored
at −20°C until used for DNA amplification. Genotypes were determined
in 5-ng genomic DNA with the Taqman allelic discrimination assay (Applied
Biosystems, Foster City, Calif). Primer and probe sequences were optimized
by using the SNP assay-by-design service of Applied Biosystems (for details,
see http://store.appliedbiosystems.com). Reactions were performed
with the Taqman Prism 7900HT 384 wells format. We used the genotype data for
each of the 2 polymorphisms to infer the haplotype alleles present in the
population by using the program PHASE, which implements a Bayesian statistical
method for reconstructing haplotypes from population genotype data.23 The alleles were defined as haplotypes such as "T-A,"
representing a thymidine (T) nucleotide at the c.454-397T>C polymorphic site and an adenosine (A) nucleotide at the c.454-351A>G polymorphic
site. Haplotype alleles were coded as haplotype numbers 1 through 4 in order
of decreasing frequency in the population (1 = T-A, 2 = C-G, 3 = C-A, and
4 = T-G).
To compare possible confounders between participants grouped by the ESR1 haplotype of interest, 1-way analysis of variance
was used for continuous variables; Pearson χ2, for dichotomous
The association between the ESR1 haplotypes
and IHD events was evaluated by stratifying participants by sex and allele
copy number (0, 1, or 2) for the haplotype of interest and using a standard
age-adjusted Cox proportional hazards model (model 1). The proportional hazards
assumption was tested and met for the Cox proportional hazards models. According
to previous analyses, we chose haplotype 1 as the risk allele.24- 27
Hazard ratios of events were computed as estimates of relative risk.
To account for possible confounding, we excluded all participants with previous
myocardial infarctions at baseline (model 2) and computed relative risks in
a multivariate model containing the following predictors of coronary heart
disease28: age, BMI, age at menopause, use
of hormone therapy, diastolic blood pressure, smoking, diabetes mellitus,
and total and HDL cholesterol levels (model 3). For the analysis of fatal
and nonfatal IHD, participants who did not have an IHD event during follow-up
were categorized as the reference group. To study the risk of experiencing
a fatal IHD event between carriers and noncarriers of the ESR1 haplotypes, a logistic regression model was used with the above-mentioned
cardiovascular risk factors as covariates.
For missing data on categorical covariates, we used a missing value
indicator, whereas for missing data on continuous covariates, we used the
median value of the respective value, as calculated from the total sample.
Missing values did not exceed 3.5% for any covariate. For all statistical
analyses, P<.05 was considered statistically significant.
All statistical analyses were performed using SPSS version 11.0.1 (SPSS Inc,
We observed the 4 possible c.454-397T>C to c.454-351A>G haplotype alleles in the following frequencies:
haplotype 1 (T-A), 53.5%; haplotype 2 (C-G), 34.7%; haplotype 3 (C-A), 11.8%;
and haplotype 4 (T-G) was present in 1 allele in 12 816 chromosomes.
Genotype distributions were in Hardy-Weinberg equilibrium.
The baseline characteristics of the study population are shown in Table 1. ESR1 haplotype
1 (c.454-397 T allele and c.454-351
A allele) was associated with diastolic blood pressure in women (P = .03).
During a mean follow-up of 7.0 years (SD, 2.0 years; range, 18 days
to 10.5 years), 1474 of the 6408 participants (23.0%) died of various causes
and 167 (2.6%) were lost to follow-up. Two-hundred eighty-five (4.4%) participants
had myocardial infarction during follow-up, of which 53 (18.6%) were fatal
and 232 (81.4%) nonfatal. Four hundred forty (6.9%) participants had an IHD
event, of which 97 (22.0%) were fatal and 343 (78.0%) nonfatal.
For the 3791 postmenopausal women in our study, ESR1 haplotype 1 was significantly associated with increased risk of myocardial
infarction, as well as IHD (Table 2).
Exclusion of 303 women with prevalent myocardial infarctions at baseline and
adjustment for age (Table 2, model
2) and subsequently for BMI, age at menopause, use of hormone therapy, diastolic
blood pressure, smoking, diabetes, and total and HDL cholesterol levels did
not significantly change the results (Table
2, model 3). Compared with noncarriers, heterozygous carriers of
haplotype 1 had 2.23 times increased risk of myocardial infarction (95% confidence
interval [CI], 1.13-4.43), whereas homozygous carriers had 2.48 times increased
risk (95% CI, 1.22-5.03). For IHD, the risk for heterozygous carriers was
increased 2.04 times (95% CI, 1.16-3.58), and for homozygous carriers, the
risk was 2.41 times higher (95% CI, 1.35-4.31). Adjustment for current use
of hormone therapy, as opposed to any previous use, did not influence the
estimates. In women, ESR1 haplotype 2 showed an opposite
but nonsignificant effect on IHD risk compared with haplotype 1: for incident
myocardial infarctions, the hazard ratio was 0.76 (95% CI, 0.55-1.05) per
copy of the C-G allele and 0.78 (95% CI, 0.59-1.01)
per allele copy for IHD. No association with myocardial infarction or IHD
was observed for the ESR1 haplotype 3 (S. C. E. S.,
unpublished data, 2003).
For the 2617 men in our study, the ESR1 haplotypes
were not significantly associated with incident myocardial infarctions or
IHD (Table 2 for haplotype 1).
Compared with noncarriers, male heterozygous carriers of haplotype 1 had 0.90
(95% CI, 0.62-1.30) times increased risk of myocardial infarction, whereas
homozygous carriers had 0.78 (95% CI, 0.51-1.21) times increased risk. Exclusion
of men with prevalent myocardial infarctions at baseline and adjustment for
age (Table 2, model 2) and subsequently
for BMI, diastolic blood pressure, smoking, diabetes, and total and HDL cholesterol
levels did not change the results (Table
2, model 3).
The 2 polymorphisms were also analyzed separately. Given the strong
linkage disequilibrium between the c.454-397T>C and 351A>G polymorphisms and the virtual nonexistence of haplotype
4, haplotype 1 fully represents the c.454-397T>C polymorphism.
Therefore, the results for haplotype 1 presented here are the same as for
the c.454-397T>C polymorphism alone. The TT-genotype group is represented by the absence of haplotype 1 and
the TC- and CC-genotypes
by the presence of 1 or 2 copies of haplotype 1, respectively. The c.454-351 A allele was nonsignificantly associated with myocardial
infarction risk in women (hazard ratio, 1.31 [95% CI, 0.96-1.80] per copy
of the A allele) and IHD (hazard ratio, 1.29 [95%
CI, 0.99-1.68] per copy of the A allele). In men,
no association was observed.
Figure 1 shows the association
between ESR1 haplotype 1 and the cumulative proportional
hazard of incident myocardial infarction during follow-up in women and men
after exclusion of participants with a prevalent myocardial infarction at
baseline and adjustment for cardiovascular risk factors. Table 3 shows that, for postmenopausal women, the effect of haplotype
1 on fatal IHD was larger than on nonfatal IHD; the hazard ratio for fatal
IHD was 6.13 (95% CI, 1.41-26.68) for homozygous genotypes and 1.86 (95% CI,
0.97-3.56) for nonfatal IHD. For the effect of haplotype 1 on fatal and nonfatal
myocardial infarctions, results were similar. For women who did not carry
haplotype 1, 13.3% of IHD events were fatal; for heterozygous carriers, this
percentage was 26.9% and was 34.0% for homozygous carriers. This resulted
in cardiovascular risk factor–adjusted odds ratios of 3.66 (95% CI,
0.56-23.80) for heterozygous and 4.39 (95% CI, 0.69-28.06) for homozygous
carriers vs noncarriers. For myocardial infarctions, the odds ratios were
similar. In men, no association with fatal or nonfatal IHD was observed.
In addition, we analyzed the first-year cumulative mortality for the
168 women who had an IHD event during follow-up. Although these results did
not reach statistical significance, Figure
2 shows that the first-year all-cause mortality in female homozygous
carriers of haplotype 1 was approximately twice that of noncarriers.
In this prospective population-based study, we observed an increased
risk of myocardial infarction in postmenopausal women who carry ESR1 haplotype 1 (c.454-397 T allele and c.454-351 A allele). The risk estimates did not change
after adjustment for clinically relevant cardiovascular risk factors, indicating
that ESR1 haplotype 1 is an independent risk factor.
Heterozygous carriers of haplotype 1 had a 2.23 times increased risk of myocardial
infarction compared with noncarriers, whereas homozygous carriers had a 2.48
times increased risk. We also analyzed the risk of IHD events by taking together
myocardial infarctions and revascularization procedures (PTCA and CABG) and
IHD mortality. With this approach, we included approximately 50% more events
and observed a similar increased risk in female carriers of haplotype 1. For
men, no association between the ESR1 haplotypes and
myocardial infarction or IHD risk was observed.
For women, the effect of haplotype 1 on fatal IHD was larger than on
nonfatal IHD, and IHD resulted in death more often in carriers of haplotype
1. Furthermore the first-year cumulative mortality after IHD was 2 times higher
for women homozygous for haplotype 1 compared with noncarriers, mainly because
of genotype-dependent differences in mortality within the first month after
an IHD event. Although the latter 2 analyses did not reach statistical significance,
these results suggest that postmenopausal women who carry ESR1 haplotype 1 have not only an increased risk of having an IHD event
but also an increased risk of death from such an event.
A number of possible indirect and direct ESR1-dependent
mechanisms through which estrogens may exert their cardioprotective effects
have been presented in the literature.29,30 Some
of the protective effects of estrogens could be mediated through systemic
effects, such as changes in lipid profile, coagulation, and fibrinolytic systems.29 However, in our study the ESR1
c.454-397T>C and c.454-351A>G genotypes were
not associated with differences in a number of cardiovascular risk factors
at baseline, such as hypertension, hypercholesterolemia, and diabetes mellitus,
which is in accordance with 2 previous studies that have shown that the c.454-397T>C and c.454-351A>G genotypes
were not associated with baseline cholesterol levels in healthy men and postmenopausal
women, as well as in men and women with preexisting coronary artery disease.31,32 Furthermore, when our analyses were
adjusted for the presence of these and other common cardiovascular risk factors,
the observed hazard ratio did not fundamentally change, which suggests that
it is not through these pathophysiologic pathways that ESR1 gene polymorphisms influence IHD disease but that this haplotype
is an independent risk factor.
Direct actions of estrogen on blood vessels could contribute substantially
to the cardioprotective effects of estrogen.29,33,34 One
of these direct actions on the blood vessel wall that may be essential in
IHD pathology is the influence of estrogen on nitric oxide production. Endothelial
cell–derived nitric oxide plays a critical role in cardiovascular disease
pathology. Nitric oxide, a primary vascular target of estrogens, not only
causes the relaxation of the vascular smooth muscle cells but also inhibits
platelet activation.35 Estrogen increases nitric
oxide production in vessels such as the aorta by increasing the expression
and enzymatic activity of the enzyme responsible for nitric oxide synthesis,
endothelial nitric oxide synthase, as well as by inducing the release of nitric
oxide. Several studies have shown that ESR1 is essential
to all 3 of these estrogen effects on vascular nitric oxide production.36- 39 This
result is further supported by the observation that basal production of endothelium-derived
nitric oxide was significantly lower in the aorta of ESR1 knockout mice compared with wild-type mice.40 Studies
have shown that ESR1 also mediates 2 other effects
of estrogen in the vessel wall: acceleration of reendothelialization and inhibition
of the vascular injury response.41,42 Particularly
the latter 2 mechanisms may explain why the ESR1 polymorphisms
are most strongly associated with fatal IHD.
How could these specific polymorphisms in the ESR1 gene influence myocardial infarction risk in postmenopausal women?
The c.454-397T>C and c.454-351A>G polymorphisms have been an important area of research in diseases
such as osteoporosis,25,43,44 cardiovascular
disease,31 and cancer,45 and
a number of hypotheses for the functional significance of these polymorphisms
have been reported in the literature. Given their location, 397 and 351 base
pairs upstream from the start of exon 2, possible functional mechanisms include
altering ESR1 expression by altered binding of transcription
factors and influencing alternative splicing of the ESR1 gene. Both of these mechanisms can be a direct result of either of
these polymorphisms or through linkage disequilibrium with a truly functional,
but so far unknown, sequence variation elsewhere in the ESR1 gene.
The first mechanism was recently supported by findings of Herrington
et al,15 which were confirmed in our own laboratory
(S. C. E. S., unpublished data, 2003). Herrington et al15 showed
that the c.454-397 T allele eliminates a functional
binding site for the transcription factor B-myb, which suggests that the presence
of this allele may result in lower ESR1 transcription.
In the presence of a decreased number of α estrogen receptors, estrogen
signaling may be less effective and, therefore, estrogen actions may be decreased.
These findings are further supported by the observation in our study
population, as well as in others, that this c.454-397 T allele has been associated with a number of phenotypes that are known
to be related to low estradiol levels and therefore low estrogen activity,
such as increased risk of osteoporosis,25 decreased
risk of osteoarthritis24 and hysterectomy,26 lower BMI,10 shorter
stature,27 and later age at menopause.26 In our study, we found the T allele
to be associated with increased risk of myocardial infarction and IHD events
in postmenopausal women, which suggests that the potentially lower ESR1 expression caused by the presence of the T allele
at the −397 polymorphic site may lead to a
higher susceptibility to IHD. The c.454-351 A allele
is also associated with IHD, which may be due to linkage disequilibrium with
the c.454-397T>C SNP or to functional significance
of the c.454-351A>G polymorphism itself.
Given the substantial differences in hormone dynamics between men and
postmenopausal women, we chose to stratify our analysis by sex. An intriguing
aspect of this study is that ESR1 haplotype 1 is
significantly associated with an increased risk of IHD in women but not men.
Similar results were also found for the associations between the ESR1 haplotype 1 with height and osteoporosis.25,27 In
these studies, the associations were also found only in women.
Recently, Shearman et al46 reported an
association between the c.454-397T>C polymorphism
and cardiovascular disease in the Framingham Heart Study. They found that
the c.454-397 T allele prevented IHD in men, which
is seemingly in contrast to our findings that the T allele
prevents IHD in women. However, there is no clear conflict with the seemingly
opposite results reported by Shearman et al,46 because
the men in our study showed a nonstatistical trend opposite to that of the
women. Furthermore, the men in the Rotterdam Study were much older than the
men in the Framingham Heart Study, and the effects of risk factors are known
to change in older cohorts. Furthermore, the Framingham Heart Study had few
cardiovascular events in women; for example, only 3 cases of myocardial infarction
were documented in the female participants of that study.
An explanation for these opposing results in men and women is not immediately
apparent. The most obvious difference between men and postmenopausal women
is the cessation of gonadal function in women. Perhaps the presence of an
intact hypothalamic-pituitary-gonadal axis in men leads to protection of the
ERa PvuII T allele in cardiovascular disease, whereas
in postmenopausal women, who are completely dependent on peripheral conversion
to estradiol, an opposing effect occurs. In men, compared with postmenopausal
women of the same age, estradiol levels are approximately 3 times higher (S.
C. E. S., unpublished data, 2003). In the presence of sufficiently high estradiol
levels, differences in ESR1 expression may not have
clinical consequences. However, estrogen deficiency after menopause in combination
with lower ESR1 expression caused by the c.454-397 T allele may lead to a higher susceptibility to IHD, which
could explain why we did not observe an association between ESR1 gene polymorphisms and IHD in men.
There are limitations to genetic-association studies. They can be influenced
by population stratification or heterogeneity, especially case-control studies
in a population of mixed racial origin. However, for our study a population-based
prospective cohort design was chosen, and all participants were of Dutch white
origin. Furthermore, the c.454-397T>C and c.454-351A>G genotypes were in Hardy-Weinberg equilibrium, and haplotype
frequencies were similar to those found in other studies of white individuals.47 Therefore, our study population may be considered
ethnically homogeneous and representative of the Dutch population. Another
limitation of association studies is the definition of the phenotype of interest
and the occurrence of phenotype heterogeneity. We diagnosed myocardial infarctions
and IHD in strict adherence to the ICD-10 guidelines
and reviewed hospital discharge letters and reports from treating specialists
to identify cardiovascular disease events. Therefore, we believe that phenotype
heterogeneity did not influence our results. Furthermore, in the Netherlands
the only way to access specialist and hospital care is by consulting a general
practitioner. Therefore, checking the general practitioners' medical records
for all participants should have resulted in a nearly complete follow-up.
As is true for all association studies, the validity of genetic-association
studies is greatly strengthened by confirmation of the results. The findings
reported in this study are supported by not only functional studies but also
associations with other phenotypes (pleiotropy) in women, such as osteoporosis,
osteoarthritis, hysterectomy, BMI, stature, and age at menopause. The presented
body of data supports the theory that in women, the presence of the c.454-397 T allele leads to lower estrogen action.
Selective nonresponse of individuals with impaired health or otherwise
at an increased risk of cardiovascular disease may have occurred. However,
such a nonresponse bias will presumably not be genotype dependent and will
not lead to overestimation of the hazard ratios.
In interpreting the clinical implications of these results, we must
consider that 78% of the population carries the ESR1 haplotype
1 risk allele and that heterozygous and homozygous carriers of haplotype 1
have a 2-fold increased risk of IHD. Perhaps we should view this not as a
"risk" allele but consider the noncarriers as having a protective allele,
which implies that noncarriers (22% of the population) have a 50% reduced
In conclusion, this population-based prospective cohort study shows
a significant 2-fold increased risk of myocardial infarction, as well IHD
events, in postmenopausal women who carry ESR1 haplotype
1 (c.454-397 T allele and c.454-351
A allele). The association was not explained by known cardiovascular
risk factors such as age, previous myocardial infarction, BMI, age at menopause,
use of hormone therapy, blood pressure, smoking, diabetes, and cholesterol
level, suggesting that haplotype 1 is an independent risk factor for IHD.
Furthermore, our results also suggest that postmenopausal women who carry ESR1 haplotype 1 have not only an increased risk of having
an IHD event but also an increased risk of death from such an event.