Introduction

It is well known and accepted that coronary heart disease (CHD) is the leading cause of death in women.[1] It also contributes a significant burden of morbidity. What still remains unclear is the role that estrogen replacement therapy (ERT) and estrogen with progestin therapy (HRT) have to play in the primary and secondary prevention of this disease complex.

Until August 1998 it was widely believed that postmenopausal ERT and HRT would positively alter clinical CHD outcomes and so should be offered to women, especially to those at highest risk for CHD.[2] This belief was based on the large body of observational data from the Nurses’ Health Study, which demonstrated that both estrogen therapy alone and estrogen with progestin were associated with a lower evidence of coronary events (relative risk [RR] 0.60 for ERT; 0.39 for HRT).[3] The burden of proof was dependent on estrogen’s beneficial effects on the lipid profile and on other parameters of cardiovascular physiology; Mendelsohn and Karas provide an excellent review.[4] In comparison, the increased risk of ERT and HRT on such negative outcomes as breast cancer appeared small (RR 1.32 for estrogen alone; RR 1.41 for estrogen plus progestin) when applied to the background risk level for a specific age group.[5] A more recent analysis is provided by Schairer and colleagues.[6]

However, there was concern that the observational data regarding CHD were skewed by a healthy-user bias;[7] that is, that the predominantly white, educated, middle-class women who chose to take ERT/HRT were in some way a healthier cohort, or possibly more careful about their health management.

In August 1998 this belief in an overall benefit of ERT/HRT was challenged by the results of the Heart and Estrogen/progestin Replacement Study (HERS), the first randomized controlled trial to compare conjugated equine estrogen (0.625 mg) and medroxyprogesterone acetate (2.5 mg) to placebo. The HERS Research Group studied 2763 older women with established heart disease and found no reduction in the overall rate of CHD events in the treatment group at the end of 4.1 years.[8] This was despite a beneficial effect on the lipid fractions (net 11% lower LDL cholesterol level and 10% higher HDL cholesterol level) in the treatment group.

Indeed, in the first year of the trial, there were more events in the treatment group than in the placebo group (57 vs 38, RR 1.52, P. 009). In addition, there was an overall increase in venous thromboembolic events (RR 2.89) and gallbladder disease (RR 1.38) in the HRT group.

Possible criticisms of the null benefit of HERS

Many criticisms were offered for these surprising results. Perhaps the culprit was medroxyprogesterone acetate. The Postmenopausal Estrogen/Progestin Interventions (PEPI) trial had shown that this progestin had blunted conjugated equine estrogen’s beneficial effect on lipids more so than did micronized progesterone.[9] However, medroxyprogesterone acetate was the progestin most used by the participants in the Nurses’ Health Study, which gave us the initial beneficial results. Careful attention to the raw data of the PEPI trial shows a very small difference in the HDL with conjugated equine estrogen plus continuous medroxyprogesterone acetate compared with conjugated equine estrogen plus micronized progesterone.

Perhaps the trial was too short, as the fourth and final year results (analyzed separately) appeared to show a trend toward benefit (years 4 and 5: 33 events in the treatment group vs 49 events in the placebo group, RR 0.67). However, analyzing the data year by year does not produce completely randomized groups, as the treatment group in the fourth year has been diminished by the clinical events occurring earlier. In comparison, the Cholesterol and Recurrent Events (CARE) trial, which compared pravastatin and a placebo in a similar population of older women with known CHD, did show a decrease in clinical events after the same time period of 4 years.[10]

Could oral conjugated equine estrogen be to blame for the greater number of deaths? Orally administered estrogens have a “first-pass effect” on liver metabolism, with consequent effects on lipoproteins and coagulation proteins. In contrast, transdermally administered 17ß-estradiol patches deliver estrogen directly into the peripheral blood stream and bypass the liver, giving rise to the postulate that they do not alter clotting factors. However, there are few studies comparing routes of estrogen administration with changes in hemodynamic parameters and with the risk of venous thromboembolic events. Numbers of participants are low and the results are variable.[11,12] Regardless of these theories, the majority of estrogen users in the observational study that provided the initial positive results—the Nurses’ Health Study—did, in fact, use oral conjugated estrogens.

Thus, we now have a problem: the only published randomized placebo-controlled trial of estrogen/progestin use and cardiovascular outcomes does not support the findings of the large body of observational evidence—evidence that appeared to make physiologic sense. Can we make physiologic sense of the HERS outcomes, which show a clear increase in thrombotic events in the venous vascular system? While estrogen in the arterial system is thought to be protective, perhaps in a susceptible cohort of women estrogen may have a procoagulant effect that outweighs (at least over 4 years) its beneficial effects on endothelial metabolism and lipoproteins. There is also the question of estrogen’s inflammatory effect. Is it pro- or anti-inflammatory in different vascular situations? With the recent report of the Estrogen Replacement and Atherosclerosis (ERA) trial (an angiographic secondary prevention trial, presented orally at the American College of Cardiology meeting in March 2000) that neither estrogen nor estrogen/ progestin slowed CHD progression, there is more and consistent support for the findings of the HERS trial.

Where does this leave the clinician in dealing with the patient who, at menopause, asks if she should take hormone therapy to decrease her risk of heart disease?

How should the evidence guide our clinical decision making?

We must base our decision making on the current best evidence. In the absence of large, well-designed randomized placebo-controlled trials (that is, prior to 1998), it was reasonable to use the results from observational studies to guide us. Equally, if the subsequent randomized trials do not support the same approach, we must heed them until there is proof otherwise.

The final analysis: What do we actually know?

We know that:

• The relationship between estrogen and CHD is more complex than initially thought.

• Estrogen has both procoagulant effects and beneficial effects on vascular endothelium and lipid levels.

• We have no evidence of benefit from HRT in the secondary prevention of heart disease. Perhaps there may even be early harm.

• These results cannot be extrapolated to primary prevention of a first event in patients who may or may not be at increased risk of heart disease.

• More evidence will be forthcoming from other angiographic trials in secondary prevention, and in two large multicentre primary prevention trials—the Women’s Health Initiative (USA) and WISDOM (UK). The Women’s Health Initiative is examining the effect of estrogen and estrogen/progestin on several diseases. Its first results on cardiac outcomes should be available for assessment by 2005.[13]

• Raloxifene (a selective estrogen receptor modulator, or SERM) has a modestly beneficial effect on lipids and is under trial for the secondary prevention of CHD. The trial is called RUTH: Raloxifene Use in the Heart.

Fortunately, failure to treat with hormone replacement therapy is not equivalent to not treating women with CHD or at high risk for CHD. The benefits of beta-blockers, ASA, and ACE-inhibitors have been established.[14-16] We know from the CARE, LIPID, and 4S trials that the HMG coenzyme A reductase inhibitors (the statin drugs) have an accepted beneficial effect on both lipid profile and CHD events in secondary prevention.[10,17,18] Pleiotropic effects at the endothelial level may account for the early benefit (at 6 to 12 months) seen in the treatment group. There do not appear to be any procoagulant effects. For the statin class of drugs, the lipid-lowering effect does appear to be a surrogate for the reduction in clinical events. The number needed to treat is at least as favorable for women as for men (CARE trial: 18.5 for women, 38 for men; 4S trial: 13.8 for women, 11 for men).

In primary prevention with statins, however, the evidence is scant. The only primary prevention trial in women is AFCAPS/TexCaps,[19] but the results must be interpreted cautiously. After an average follow-up of 5.2 years, the relative risk reduction for CHD events in women was reported as a dramatic 54%. However, the event rates were low in both the placebo and the treatment groups, with an absolute risk reduction of only 1.2%. The number needed to treat for women is 83.3.

Clearly, in primary prevention, trials will need to continue far longer to see if there is long-term benefit in clinical outcomes without other serious side effects. Longer hormone replacement/CHD trials may, however, still be complicated by an early increase in procoagulant events.

What should be considered in a risk factor assessment? 

Most importantly, we must remember that hormonal status is but one risk factor for CHD. Cigarette smoking, diabetes, and hypertension are all strongly correlated with increased risk of CHD. Recognizing and managing these conditions has been shown to be cost-effective.[20] Obesity and lack of exercise are also factors that predispose patients to CHD or modify the risk profile for CHD.

As well as considering our patients’ risks for atherosclerotic events, we should address their risks for breast cancer, other cancers, osteoporosis, and Alzheimer’s disease. We need to discuss each patient’s genetic predisposition, environmental factors, lifestyle behaviors, and belief system. This will lead to a risk-benefit analysis specific for each patient.

What is the most prudent course?

We should be frank with patients about our level of knowledge: what we know, what we surmise, what information is on the horizon. In women with known CHD, there is no evidence to support starting HRT/ERT. Indeed, there may be early harm. However, for those women in the secondary prevention group who have been on hormone replacement for longer than 1 year, there is also no evidence that their medication should be stopped. The HERS study suggests that the greatest burden of events occurs in the first year of treatment, perhaps selecting out those women most susceptible to thrombotic events.

Women without established CHD make up the largest group seeking treatment. Do women in this primary prevention (prevention of the first event) group differ in their physiology from women who have already had an event? We don’t know. It has been shown, at least by animal studies, that estrogen requires an intact endothelium to prevent future damage.[21] What about women over the age of 60 years who may already have experienced a silent infarct?

The most prudent course is: “first do no harm.” Identify and treat all risk factors for CHD. In women who have known coronary heart disease or cerebrovascular disease, avoid starting ERT/HRT. For women free of these diseases, offer hormone replacement therapy (estrogen alone or estrogen/ progestin) as indicated, for the short-term management of postmenopausal symptoms. For long-term disease prevention, ERT/HRT is a good choice for those women at high risk for osteoporosis. There is no clear direction for women at high risk for atheromatous disease or with diabetes mellitus. For safe and effective prevention of coronary heart disease with ERT/HRT, both primary and secondary, I am waiting for the evidence.

Addenda 

Since this article was accepted for publication, four events have occurred that deserve comment.

First, the ERA study has been published, and the data are available for examination.[22] After a mean of 3.2 +/- 0.6 years, there was no difference in the progression of CHD between the treatment and the placebo groups. There was also no evidence of excess early events as in the HERS trial. However, angiographic data need to be interpreted with some caution, as they may not be fully indicative of what is actually happening within the vessel wall. Vessel wall ultrasound may give a better answer.

Second, the safety board for the Women’s Health Initiative notified the participants of a small but definite increase in heart attack rates in the treatment group. The effect was not large enough, however, to indicate a need to discontinue the trial.

Third, in July 2001 the investigators in the Nurses’ Health Study published their observational data from the cohort of women in the secondary prevention of coronary heart disease category. Their objective was to provide additional information on a time trend in risk as observed in HERS. The follow-up was up to 20 years. For short-term current users (defined as less than 1 year of hormone use after the initial coronary event, regardless of hormone use before the event), the multivariate adjusted relative risk of a second coronary event was 1.25 compared to women who had never used hormones. After longer-term use (defined as more than 2 years), the similarly calculated relative risk was 0.38. An alternative analysis, in which short-term use was largely limited to women who first began hormone therapy after their initial coronary event, showed an even higher risk of recurrence (RR 2.10).[23]

Fourth, results have been presented from the first two randomized controlled trials focusing on secondary prevention regarding postmenopausal HRT use and the risk of stroke (HERS and WEST). Previous observational studies gave inconsistent answers. HERS had as a prespecified secondary outcome the effect of postmenopausal HRT on the risk of stroke and transient ischemic attack.[24] The data show that conjugated equine estrogen at 0.625 mg with medroxyprogesterone acetate at 2.5 mg daily had no significant effect on stroke risk over a period of 4.1 years in a population of women with known CHD. In contrast to the HERS CHD data, there was no “time trend to benefit” (although this latter benefit is debatable).

Similarly, the Women’s Estrogen for Stroke Trial (WEST) results show that 17ß-estradiol at 1 mg daily postmenopausally did not decrease the incidence of recurrent stroke or all-cause mortality over 2.5 years in women with known cerebrovascular disease.[25]

The pathophysiology of stroke is even more complex than that of CHD. It is not surprising, therefore, that the effects of HRT on cerebrovascular disease are equally complex and may appear contradictory in different populations of women.[26]

Two questions remain:

• Is secondary prevention of CHD with HRT appropriate for some women and not for others?

• Are primary and secondary prevention of CHD with HRT different entities?

Another few years (2005) is not too long to wait for answers to these questions.

Darlene M.S. Hammell, MD

Dr Hammell is a clinical assistant professor in the Department of Family Practice at the University of British Columbia, active staff in the Capital Health Region, Victoria, a consultant at the Victoria Lipid Clinic, and a member of the Working Group on Dyslipidemia, Health Canada.