Hormone Replacement Therapy and the Cardiovascular System: Lessons Learned
and Unanswered Questions

Pamela Ouyang, MB, BS, FACC; Erin D. Michos, MD; Richard H. Karas, MD,
PhD, FACC .J Am Coll Cardiol. 2006;47(9):1741-1753.©2006 Elsevier Science, Inc.

Abstract

Cardiovascular disease is the leading cause of death among women in the
U.S., exceeding breast cancer mortality in women of all ages. Women
present with cardiovascular disease a decade after men, and this has been
attributed to the protective effect of female ovarian sex hormones that is
lost after menopause. Animal and observational studies have shown
beneficial effects of hormone therapy when it is initiated early in the
perimenopausal period or before the development of significant
atherosclerosis. However, randomized, placebo-controlled trials in older
women have not shown any benefit in either primary prevention or secondary
prevention of cardiovascular events, with a concerning trend toward harm.
This review outlines the lessons learned from the basic science, animal,
observational, and randomized trials, and then summarizes yet-unanswered
questions of hormone therapy and cardiovascular risk.

Introduction

Cardiovascular disease is the leading cause of death among women in the
U.S., accounting for more than 500,000 deaths annually.[1] Mortality
attributable to coronary heart disease (CHD) exceeds breast cancer
mortality in women at all ages.[2] The vast majority of these
cardiovascular events occur in postmenopausal women. Heart disease
develops in women on average 10 years later in life compared with men, and
this lag has been attributed to the protective effects of female sex
hormones, particularly estrogens, before menopause.[3]
Initially, data from animal studies and observational studies such as the
Nurses’ Health Study[4] strongly supported a protective cardiovascular
benefit of hormone therapy (HT) after menopause, an effect not supported
by randomized placebo-controlled trials in both secondary prevention[5]
and primary prevention,[6] which instead showed a concerning trend toward
harm. However, many unanswered questions remain.
This review outlines lessons learned from basic science of estrogen action
and animal studies and from observational and randomized trials, followed
by a discussion of as-yet-unanswered questions about HT and cardiovascular
risk.

Molecular and Cellular Basis of Estrogen in Vascular Biology

Estrogen can have both positive and negative effects on the cardiovascular
system[7] (Fig. 1). On the positive side, estrogen has potentially
beneficial effects on lipid parameters, such as reducing low-density
lipoprotein cholesterol (LDLC) and increasing high-density lipoprotein
cholesterol (HDLC), facilitating nitric oxide-mediated vasodilation, and
inhibiting the response of blood vessels to injury and the development of
atherosclerosis.[8] On the negative side, estrogens increase
triglycerides[9,10] and inflammatory markers such as C-reactive protein
(CRP).[11,12] Estrogen also has many prothrombotic effects, such as
increasing circulating levels of prothrombin and decreasing antithrombin
III,[13,14] contributing to an increased risk of venous thromboembolic
events. Importantly, many of these effects of estrogen are mediated by
first-pass effects on the liver, and thus result from oral but not
transdermal administration of HT. For example, increased levels of CRP
seem to occur only with oral estrogen administration. The extent to which
this is associated with an increase in cardiovascular disease risk is
uncertain.[15,16] These observations underscore the potential importance
of the mode of administration on the overall effects of HT on CHD risk as
discussed below.

Mechanisms of Estrogen Action

As our understanding of the mechanisms by which estrogens affect the
cardiovascular system has increased, it has become clear that the
complexity of the biological effects of estrogens are reflective of quite
complex mechanisms of action.[7,8] As noted above, estrogens regulate a
variety of systemic or circulating factors, including lipids, inflammatory
factors, and members of the coagulation/fibrinolytic cascades. Estrogens
can also act directly on the heart and the vasculature. The effects of
estrogen are mediated by estrogen receptors, of which two are known, ERα

And ERß, and both are expressed in cardiovascular cells and tissues[7,8]
Estrogen receptors are classically thought of as ligand-activated
transcription factors that reside in the cell nucleus and regulate gene
expression in response to hormone binding. This mechanism, often referred
to as the genomic pathway, likely underlies the longer-term effects of
estrogen, such as those on circulating levels of lipids and coagulation
factors. More recently it has become clear that estrogen receptors also
transduce the rapid effects of estrogen that occur within minutes[17] and
are referred to as non-genomic because they do not depend on changes in
gene expression.

These rapid effects of estrogen are mediated by a
subpopulation of estrogen receptors localized to cell membrane signaling
domains called caveolae. The best-studied example of this non-genomic
pathway for estrogen action in the cardiovascular system is the activation
of endothelial cell nitric oxide synthase that results in arterial
vasodilation in response to acute administration of estrogen.
Impact of Disease State on the Cardiovascular Effects of Estrogen
There is growing evidence showing that the effects of estrogen on the
vasculature depend in part on the extent to which atherosclerosis has
become established. For example, estrogen receptor expression is markedly
diminished in atherosclerotic arteries,[18,19] and thus, to the extent
that direct, receptor-dependent effects on the vasculature contribute to
the potential for anti-atherosclerotic effects, this will be diminished or
absent in diseased arteries. In addition, the effects of estrogen on a
given pathway may have different consequences depending on the state of
health of the underlying vessel.For example, estrogen up-regulates
specific members of the matrix metalloproteinase (MMP) family such as
MMP-9.[20] The MMPs degrade the extracellular matrix with the arterial
wall. Thus, in a non-diseased artery, an estrogen-induced increase in
MMP-9 may have little or no consequences, whereas in an atherosclerotic
artery, where MMP-9 is expressed in the shoulder region of an
atherosclerotic plaque, an increase in MMP-9 activity could conceivably be
associated with an increased risk of plaque rupture and thus acute
coronary syndromes.

The Fitzgerald laboratory has recently shown that
estrogen-mediated up-regulation of cyclooxygenase-2 plays an important
role in retarding atherosclerosis in a hypercholesterolemic mouse
model.[21] This suggests that atherosclerotic arteries with impaired
cyclooxygenase-2 responses may also lose this potentially beneficial
effect of hormone treatment.

More direct support for the hypothesis that the effects of estrogen on
cardiovascular risk depend on the timing of initiation of therapy in
relation to the extent of underlying atherosclerosis comes from the
Clarkson laboratory. Using a well-established monkey model of
atherosclerosis, Clarkson and colleagues have shown that the
antiatherosclerotic effects of oral conjugated equine estrogens (CEE) are
apparent only in monkeys with minimal underlying atherosclerosis at the
time that therapy is initiated, as reviewed in Karas and Clarkson,[20] a
finding also supported by rodent and rabbit studies.[22,23]

Cardiovascular Disease and Menopause

Interest in the role estrogen plays in cardiovascular disease was
stimulated by the observed increase in cardiovascular events after
menopause. In 1976 the Framingham investigators reported a 2.6-fold higher
incidence of cardiovascular events in age-matched postmenopausal women
compared with premenopausal women.[24] The excess CHD risk associated with

Surgical menopause was 2.7-fold higher compared with premenopausal women
of the same age (p < 0.01)[25] and 2.2-fold higher compared with women
with a natural menopause. This excess risk seemed to be prevented by
estrogen replacement therapy.[26] Plasma lipoproteins were thought to play
a role in the increased CHD risk that menopause confers because total
cholesterol, LDL cholesterol, and triglyceride levels all increase in
women after menopause,[27] and HT seemed to counter these unfavorable
effects of lipids, although the cardioprotective HDL cholesterol levels
also decreased.[28] In addition, there is an age-associated increase in
the incidence of cardiovascular disease for both premenopausal and
postmenopausal women independent of the effects of HT. Overall, however,
data indicate that withdrawal of estrogen during menopause is associated
with an increased risk of heart disease above that seen for premenopausal
women. This led to interest in the potential cardiovascular benefit from
postmenopausal estrogen replacement therapy.

Animal Studies of Hormone Replacement

In animal studies, estrogens exert vasodilator,[29] anti-inflammatory,[30]
and antiatherosclerotic[31] properties, as well as favorably affecting
lipid profiles. In a study of randomized ovariectomized
hypercholesterolemic rabbits, estradiol significantly reduced
atherosclerosis progression compared with levonorgestrel or no
hormones.[32] A series of studies have also shown that estradiol
significantly lessens the response to vascular injury in mice and further
implicate ERα as the specific estrogen receptor that mediates this
vasculoprotective effect.[33-36] Similarly in ovariectomized monkeys,
17ß-estradiol or CEE reduced coronary artery atherosclerosis compared with
control animals by 50% (p ≤ 0.05)[37] to 72% (p < 0.04).[38] Although the role of estrogen replacement seemed promising in the animal studies, the data regarding progesterone were more conflicting.[39]

Observational Trials of Hormone Replacement

Overall, the animal studies suggested a promising role of estrogen
replacement after menopause. Simultaneously, a series of observational and
case-control trials also suggested benefit (reviewed in Table 1 ). The
majority of the smaller case-control studies[40-45] showed nonsignificant
trends toward reduction in CHD events with overall odds ratios ranging
from 0.69 to 0.9. However, a large case-control study did show a
significant association with HT and reduced incidence of first myocardial
infarction (MI).[46] Longer duration of use seemed to confer even more
cardiovascular benefit.[47]

In addition to reducing CHD events, cross-sectional data suggested less
subclinical atherosclerosis in HT users.[48] In an observational analysis
of the Cardiovascular Health Study of women >65 years of age, estrogen
users had lower levels of subclinical disease as measured by a variety of
surrogate end points.[49] Even more promising, the much larger
observational Nurses’ Health Study of 70,000 asymptomatic women showed a
lower incidence of CHD events and all-cause mortality in HT users compared
with nonusers.[4,50,51] It is important to note that most women in the
Nurses’ Health Study likely started taking HT in the perimenopausal period
and were free of known CHD at the start of the study.
Randomized Trials of Hormone Replacement Therapy
The animal and observational studies were followed by a series of
randomized, placebo-controlled trials of HT in both primary and secondary
prevention, and with both surrogate and cardiovascular event outcomes,
which failed to confirm cardiovascular benefit (reviewed in Table 2 ).

Although two smaller randomized trials using the surrogate end points of
carotid intima-medial thickness (IMT)[52] and brachial reactivity[53]
favored estrogen, three other randomized trials looking at atherosclerotic
progression by coronary angiography showed no benefit.[54-56] In women who
already had coronary disease[57] or increased subclinical atherosclerosis
on carotid IMT assessment,[58,59] HT had no impact on disease progression.
Also in a study of postmenopausal women presenting with unstable angina,
acute HT started in the hospital setting had no effect on reducing further
ischemic events evaluated by ambulatory electrocardiographic
monitoring.[60] Similarly, HT given to women with recent acute stroke did
not reduce subsequent stroke or mortality.[61]

The Heart and Estrogen/Progestin Replacement Study (HERS) was the first
published secondary prevention trial in 2,763 women with known CHD
followed up for the primary outcome of the cardiovascular events of
nonfatal MI or CHD death. At a mean of 4.1 years, there was no significant
difference in the HT arm versus the placebo arm (hazard ratio of 0.99, 95%
confidence interval [CI] 0.80 to 1.22).[5] There seemed to be a trend
toward benefit at longer durations of therapy, but this was not supported
with the longer follow-up period of seven years (hazard ratio of 0.99, 95%
CI 0.81 to 1.22).[62]

The largest randomized clinical trial is the Women’s Health Initiative
(WHI), which included 16,608 women with an intact uterus randomized to CEE
and medroxyprogesterone acetate (MPA) or placebo with a 5.6-year mean
follow-up.[6] This was stopped early because of an increase in breast
cancer among HT users with no cardiovascular benefit. An additional 10,739
women with hysterectomy were randomized to CEE alone or placebo, and that
trial also showed no cardiovascular benefit.[63] The findings for the
combined CEE-MPA arm suggested that for 10,000 person-years there would be seven more CHD events, eight more strokes, eight more pulmonary embolisms, and eight more invasive breast cancers. For the CEE alone group, there would be an absolute excess of 12 strokes per 10,000 people despite a risk reduction of 6 fewer hip fractures. The WISDOM trial,[64] based in the
United Kingdom with a design similar to the WHI trial, was discontinued in
2002 after the results of the WHI trial were published.

Randomized placebo-controlled trials reduce many of the biases inherent in
observational studies but have other limitations. The WHI trial is a very
important study that has changed the national view of HT. Although
primarily a primary prevention trial, a small number of patients did have
established disease, including a history of MI (1.6%), history of angina
(2.8%), history of coronary bypass surgery/percutaneous coronary
intervention (1.1%), history of stroke (0.7%), and/or CHD risk factors of
diabetes (4.4%), hypertension on therapy (35.7%), and hyperlipidemia on
therapy (12.5%). The mean age of 63 years puts the majority of these women
at least 10 years after menopause at the time of initiation of HT. There
was also significant crossover between the two arms. Of consideration,
only one drug regimen, CEE 0.625/MPA 2.5 mg orally per day, was tested, so
these findings may or may not apply to lower dosages of these drugs, other
formulations, or other routes of delivery.

Questions to be Answered

Given the findings of the HERS and of WHI trials, is the HT discussion
ended for good? Hardly. The conflicting results from animal/observational
studies compared with the randomized controlled trials raise many
unanswered questions. These include whether some of the discrepancy is
related to the age of the women in the studies, the timing of initiation
(perimenopausal or postmenopausal), the amount of atherosclerosis at the
time of initiation (primary vs. secondary prevention), the dosage, and the
preparation form (transdermal, oral, or intravenous with or without
progesterone), and whether there are genetic aspects to benefit or harm
from HT.

Recently, Prentice et al.[65] reanalyzed the data from the observational
trials adjusting for time from estrogen-plus-progestin initiation and
confounding variables, and found that the readjusted hazard ratio
estimates between the observational and experimental trials became much
more similar for outcomes of CHD and thromboembolism, although less so for
stroke.[65] These analyses suggest that the apparent discrepancies between
clinical trial and observational study findings may be substantially
explained by classical confounding and differences in distributions of
time of initiation.

Timing of Initiation

Because atherosclerosis accelerates after estrogen deficiency, it would
seem logical that estrogen replacement would have the most benefit when
starting early in perimenopausal women. Most women in the observational
trials such as the Nurses’ Health Study, which suggested a protective
effect of estrogen, started HT during the perimenopausal transition,[66]
whereas the WHI trial contained too few women in the perimenopausal period
to evaluate whether any cardiac protection was seen. In the WHI trial the
average age was 10 years after menopause, an age at which subclinical
atherosclerosis has developed in many women.[67]

In support of the notion that timing of initiation is critical, animal
studies also showed no benefit of estrogen in animals that already had
artery damage, either from balloon injury or from atherosclerotic diet,
before initiation of HT.[68] These animal studies are consistent with the
findings of the secondary prevention (i.e., HERS) trials.

In postmenopausal women in the Cardiovascular Health Study, estrogen replacement only caused vasodilation of the brachial artery in younger
women without clinical or subclinical cardiovascular disease, suggesting
that the favorable effects of estrogen may be limited to only those in
whom atherosclerotic vascular disease has not yet developed.[69]

The Kronos Early Estrogen Prevention Study (KEEPS) is a multicenter
randomized placebo-controlled clinical trial that will evaluate the
effectiveness of 0.45 mg CEE or 50 µg transdermal estradiol (in
combination with 200 mg progesterone) in preventing the progression of
carotid IMT or coronary calcium in women who are within 36 months of their
final menstrual period.[70] It is hoped that the KEEPS trial will provide
some answers to the important question of whether HT will have a
beneficial role if started early, although a relatively small sample size
and the use of a surrogate end point represent limitations of this study.
Hodis et al.[71] have recently launched the Early vs. Late Intervention
Trial with Estrogen (ELITE), which is also focused on examining the
potential importance of time since menopause on the cardiovascular effects
of HT. In the ELITE study, the effects of oral 17ß-estradiol on carotid
IMT will be compared directly in perimenopausal women versus those >6
years after menopause.

Dose

Although 0.625 mg CEE clearly showed no cardiovascular benefit in the HERS
and WHI trials, the observational Nurses’ Health Study found the
protective effect of CEE only in the lower doses of 0.3 mg and 0.625 mg,
whereas 1.25 mg and higher doses were not protective. In a small
randomized, double-blind crossover trial by Koh et al.[13] of 57
postmenopausal women on progesterone, lower-dose CEE 0.3 mg compared with 0.625 mg had similar favorable effects on HDL, triglycerides, and brachial
reactivity, but had fewer prothrombotic effects and a smaller increase in
CRP. In postmenopausal diabetic women without a recent MI among the Kaiser
Permanente database, low-dose or medium-dose estrogen (<0.625 mg)
decreased the risk of MI, which was not seen with a higher dose.[72]
Whether a lower dose of estrogen such as 0.3 mg CEE would provide
cardioprotection without increasing thromboembolism remains to be seen.

Route of Delivery

The formulation of estrogen used in the large clinical trials and in the
majority of the smaller studies was CEE with or without MPA. There are
extremely limited randomized trial data for other preparations of HT.
Transdermal estrogen delivery provides sustained release of estrogens and
more constant blood levels than oral administration. When estrogen is
given orally, it has first-pass effects on the liver. Transdermal
preparations avoid the first-pass effects on the liver and have less
effect on the lipoprotein profile.[73] Estradiol-to-estrone conversion is
slower in parental administration, but transdermal delivery more commonly
facilitates an estradiol-estrone ratio of about 1, which is similar to the
physiological ratio in the pre-menopausal state.

Other differences that favor the transdermal approach include a neutral
effect on CRP, a decrease in factor VII and fibrinogen, and a reduction in
blood pressure.[74] Because the first pass through the liver is avoided,
there may be less induction of a prothrombotic state with the transdermal
preparation. The Estrogen and Thromboembolism Risk study Group (ESTHER)
case-controlled trial found that oral estrogen but not a transdermal
formulation increased the risk of venothromboembolism in postmenopausal
women on HT compared with control subjects.[75]

In a case-controlled study, transdermal users seemed to have similar level
of cardioprotective effects as those receiving oral preparations.[46]
These effects are not significant, but this may be attributable to small
sample size. However, an animal study using transdermal estradiol did not
find inhibition of aortic atherosclerosis.[76]

Overall, the benefits of a transdermal estrogen preparation over an oral
one seem encouraging, but further randomized trials are warranted.[77] It
is hoped that the KEEPS trial,[70] which will randomize healthy
perimenopausal women to oral versus transdermal hormone replacement, will
provide information on this important question.

Genetics

It is possible that genetically determined subgroups of women may benefit
by or be harmed from HT. Studies have shown that the cardiovascular
effects of HT differ in individuals with specific genetic variants of
certain genes such as apolipoprotein E4 (-) and myeloperoxidase.[78,79]
Recent work has also shown that genetic variation in the estrogen receptor
itself may modulate the cardiovascular effects of HT, and also alter the
underlying risk of CHD. The Herrington laboratory recently showed that a
specific genetic variant of ERα is associated with an enhanced
HDL-increasing effect of HT.[80] Quite surprisingly, this same genetic
variant of ERα was also recently shown to be associated with an
approximately three-fold increased risk of MI in men in the Framingham
heart study.[81]

Statin Use

Perhaps concomitant statin use may attenuate the negative cardiovascular
effects of HT. Subgroup analysis of the HERS trial showed that the
increased cardiovascular risks of HT in this population of women with
established CHD did not occur in women taking statin therapy; however,
there was no incremental risk reduction for cardiovascular events in women
on both statin and HT compared with statin alone.[82]

Summary

In conclusion, the HT controversy has not yet been laid to rest. Current
randomized clinical trial data support the American Heart
Association/American College of Cardiology guidelines that HT should not
be prescribed for prevention of cardiovascular disease.[83,84] However, it
remains possible that some formulations and doses of HT may have favorable
cardiovascular benefits when initiated earlier in the pre-menopausal or
perimenopausal period in women without pre-existing atherosclerotic
disease. We await the KEEPS and ELITE trial results, among others, to
further answer this important issue.