Hormone Replacement: Some Surprising Bad News

Abstract & Commentary

Synopsis: The late trend toward a decrease in CHD events may be valid, perhaps as a result of the favorable lipid effects that induced benefit after 36-48 months.

Source: Hulley S. JAMA 1998;280:605-613.

Conventional wisdom, based on a variety of small trials and many observational studies as well as meta-analyses, has suggested that hormone replacement in post-menopausal women results in approximate halving of coronary disease rates. No large, prospective randomized trial has ever been reported until the Randomized Trial of Estrogen plus Progestin in Secondary Prevention of Coronary Heart Disease in Post Menopausal Women (HERS) (Hulley S. JAMA 1998;280:605). The results of this long-awaited randomized trial of estrogen plus progestin (HRT) for secondary prevention of coronary heart disease in post-menopausal women sent shock waves around the world, as the results did not confirm the widely held belief that estrogen replacement would be protective against coronary heart disease (CHD). Twenty centers enrolled 2763 post-menopausal women with established CHD, beginning in 1993 and ending in early 1998. Participants had a mean age of 67; 90% were white. All underwent careful screening, and, during this study itself, the women had periodic breast and uterine examinations as well as assessment of clinical end points. Treatment consisted of conjugated equine estrogen, 0.625 mg and medroxyprogesterone (MPA) 2.5 mg, or placebo; follow-up averaged 4.1 years. Approximately 82% of those assigned to hormones were taking therapy at the end of year 1 and 75% by year 3. The primary end point was nonfatal MI or CHD death. A variety of secondary cardiovascular outcomes were assessed, as were total mortality, cancer, deep venous thrombosis, etc. Baseline lipids were abnormal, with mean LDL cholesterol of 145 and HDL of 50. The mean time since last menstrual period was 18 years. More than half of the population was overweight; less than 40% exercised regularly. Approximately 25% had been exposed to post-menopausal estrogen in the past.

The results reveal no difference between placebo and HRT in the primary end point. Furthermore, there was an actual increase in CHD events during the first 24 months of the study, compared to placebo, with decreasing CHD events vs. placebo in the last two years of the study. The risk ratio for CHD death or nonfatal MI with hormones was 1.5 in year 1, falling to 0.7 in year 3, and 0.67 in years 4 and 5. The most dangerous time for patients was in the first four months, when the risk hazard was 2.3, particularly for nonfatal MI. There was no difference in total mortality or cancer deaths. There was an almost three-fold greater likelihood of deep venous thrombosis in the HRT cohort, and there were 11 vs. four episodes of pulmonary emboli. Gallbladder disease was somewhat more common in the HRT cohort. Plasma lipids demonstrated a decrease in LDL of 14% by year 1 and an increase in HDL by 8%. These changes remained consistent throughout the trial.

Hulley and associates speculate that the late trend toward a decrease in CHD events may be valid, perhaps as a result of the favorable lipid effects that induced benefit after 36-48 months. Furthermore, they speculate that the early hazard of HRT could be due to a pro-thrombotic effect of hormone therapy. They question whether a different progestin might have produced different results. Finally, they emphasize that the HERS Trial is substantially different from previously reported hormone studies by its duration, the randomized placebo-controlled nature, and studying women primarily for CHD events.

Comment by Jonathan Abrams, MD

The HERS Trial is a disappointment to the many who believed that post-menopausal estrogen replacement would have a significant favorable effect on coronary artery disease. Perhaps the most important message is stated by Petitti, who wrote an accompanying editorial and concluded "commitment to randomized trials as the standard of proof must be especially strong when the public health implications are so great." Thus, in spite of a large amount of observational "evidence" and the collective wisdom of many "experts" (including myself, Clin Cardiol Alert, 1998;21:218-222). The results do not confirm a favorable effect of HRT in older post-menopausal women with established CHD.

There have already been commentaries about this trial and speculations as to why HERS was not positive. Some commentators believe that the early adverse trend within the first 12-24 months (reversing to a favorable trend in the last two years of the study) implies that the duration of the HERS study was not long enough and that a longer period of HRT might demonstrate a favorable effect on CHD morbidity and mortality. Some suggest that MPA is the culprit, attenuating and perhaps canceling out the positive effects of estrogen; certainly, the literature is in conflict regarding whether progesterones do or do not add additional protection to estrogen with respect to the vessel wall. Hulley et al and Petitti speculate that the pro-thrombotic effects of the hormone combination were particularly active in the early years of the study, resulting not only in increased coronary events but also in a higher rate of deep venous thrombosis, particularly in year 1. Down-regulation of estrogen receptors by MPA is a possibility. Furthermore, it is possible that a different progestin might have resulted in a more favorable outcome. Whatever the reasons, the HERS data stand on their own, representing a well-conducted, double-blind placebo controlled trial that was unable to show benefit of combination hormone replacement in spite of favorable lipid effects over the course of the study. Another factor could well be that the study women were older, averaging 67 years at entry and 18 years post-menopausal before beginning HRT. Thus, the extent of the existing coronary vascular disease might have been substantial and less amenable to favorable effects of HRT over a relatively short period than might be true in younger women.

There has always been considerable emphasis that women who use hormones in the post-menopausal years may be different than those who do not. Many demographic characteristics, including socio-economic class, lifestyle, and general health, have been shown to be more favorable in HRT users. The concepts of prevention bias and compliance bias are germane to this discussion, implying that HRT users are healthier in many ways and are more likely to be compliant with their medications. Both these factors are impossible to unravel from the available observational data, but they clearly imply that HRT women might have a substantially better prognosis unrelated to the administration of estrogens.

It should not be forgotten that the vascular biology story regarding estrogen is extremely positive. The conventional wisdom that HRT is beneficial to the cardiovascular system is not disproven by HERS, but, certainly, the hypothesis has suffered a serious setback. Nevertheless, use of HRT for post-menopausal symptoms, uro-gynecologic problems, osteoporosis, and possible slowing of dementia is well founded. At the present time, there is no mandate for women with vascular disease to be placed on hormones in hopes that their overall CHD risk will be reduced, although this is still a possibility, as suggested by the last 24-month experience of the HERS Trial. Few adverse outcomes were noted in this trial other than the expected increase in venous thrombosis and a trend toward more pulmonary emboli, which abated over the last two years. Thus, physicians are urged to carefully counsel post-menopausal women with vascular disease regarding HRT, providing a full list of risks and benefits. Such treatment is appropriate for many, if not most, women who wish to prevent one or more of the adverse sequalae of the menopause.

Whether younger healthy women at risk for CHD should be placed on HRT, or perhaps estrogen alone, remains an unresolved question. It may be that use of hormones relatively soon after menopause might be critical with respect to slowing down the atherosclorotic process. Unfortunately, data are not available and may not be for some time. The Women’s Health Initiative is addressing this question, but completion is not scheduled until the year 2005. Finally, the question of whether a progestin should be used in conjunction with estrogen replacement is unresolved. There are data suggesting that MPA may not be the ideal compound. Protection of the uterus from unopposed estrogen remains a critical problem in HRT; many to most women today are given both compounds. While HERS does not provide a smoking gun to discard the protective effects of progestin, it does give considerable stimulus to the search for a potentially better progestin than MPA itself.