Hormone Replacement Therapy: The Last Waltz or a New Step?

Two years ago, the largest placebo-controlled trial of hormone replacement therapy (HRT) was shut down when women receiving treatment showed an increased risk of heart disease and breast cancer. In this special Crossroads essay, two physicians who study the effect of hormones on the development of mood disorders explain why they think the outright dismissal of HRT was hasty and suggest that researchers not abandon their investigations of HRT.

Two years ago, the publication of the results of the Women's Health Initiative (WHI) turned our attitudes about HRT on end. The longitudinal study was designed to test two widely prescribed forms of HRT: 1) Premarin, a mixture of estrogens extracted from horse urine and 2) Prempro, which contains both Premarin and Provera, a "progestin," or synthetic progesterone-like compound. Because estrogen when administered on its own can trigger a buildup of the uterine lining that acts as a precursor for endometrial cancer, Premarin alone was given only to women who'd had hysterectomies. Most forms of HRT include a progestin such as Provera to avoid the uterine effects caused by estrogen alone. WHI selected Prempro because it is the most commonly prescribed form of HRT.

The studies examined the effects of HRT on breast and colon cancers, bone fractures, and cardiovascular diseases, including coronary heart disease, stroke, and blood clots. Before the WHI, HRT was prescribed to relieve the acute symptoms associated with menopause--such as hot flashes, vaginal dryness, and depression--and to prevent cardiovascular disease and bone fractures. Prescribing HRT to prevent these disorders represented a calculated risk: HRT had previously been associated with a decrease in cardiovascular deaths by up to 50%, but it also increased the risk of breast cancer by 30% after 4 years of treatment. Because cardiovascular disease accounts for 50% of female deaths and breast cancer accounts for only 4%, the benefit provided by HRT seemed worth the risk.

Not so, concluded the WHI study, which was prematurely terminated in July 2002 because women receiving HRT showed an increased--not decreased--incidence of coronary heart disease and an increased incidence of breast cancer compared with those taking a placebo (see "Hot Flash"). With this and subsequent reports, HRT was basically left for dead. "The latest decision by the overseers of the WHI study disproves for good the theory that replacing the sex hormones that fall steeply after a woman reaches the menopause would yield an overall benefit to her health. It marks the final collapse of a medical and commercial edifice that had been constructed over three decades," declared The Washington Post. The journal Science agreed: "This is the final nail in the coffin for hormone replacement therapy." Clearly, the idea that HRT is good for everyone is incorrect. But is the now-prevailing view that HRT is bad for everyone correct? A closer examination of the data suggests that we might have been too quick to pronounce HRT dangerous and valueless.

As the quote from The Washington Post suggests, the original rationale for HRT was to reverse or prevent the dramatic decrease in sex hormone quantities that occurs in the year after the last menses. This drop was believed to contribute to the rise in heart disease and osteoporosis in women after menopause. The hypothesis that HRT prevents subsequent health problems, then, should be tested in women who are nearing menopause, that is, about 51 years old (the mean age of menopause in the United States). In contrast, the mean age of the women in the WHI study was 63.3, with an age range of 50 to 79 and two-thirds of the subjects over 60. Only 10% of the participants were age 50 to 54.

This paucity of perimenopausal women could be problematic for two reasons. First, animal studies suggest that there is a "window of prevention," after which administration of estrogen will not show a beneficial health effect. Monkeys given estrogen (Premarin) within 2 months of experiencing an artificial menopause induced by removal of their ovaries do not show the degeneration of the coronary arteries that otherwise follows loss of these organs. This protective effect is not seen, however, if estrogen administration is delayed for 2 years following the surgery. If most of the volunteers in the human studies were more than a year or two into menopause, the window might have closed. Second, older women are less healthy than younger women, and estrogen might have very different effects in the presence of disease. Among the "apparently healthy" postmenopausal women participating in the WHI, 7.7% had a history of prior cardiac disease, 36% had received treatment for high blood pressure, 49% were current or former smokers, and 34% were obese--not a picture of perfect health, but fairly representative of a population of women in their 60s and 70s. In addition to their known health problems, older women are far more likely to have undetected cardiovascular disease. For example, women in their 60s are more likely to have coronary arteries lined with fatty lesions that can obstruct blood flow. Estrogen can increase the risk for heart attack or stroke by turning on an enzyme that can chew away the membrane that contains these lesions, causing them to rupture.

The presence of preexisting disease is one potential explanation for the increased risk of coronary heart disease during the first year of taking HRT, seen in both the WHI and HERS (Heart and Estrogen/progestin Replacement Study) studies. The HERS study tested whether HRT would prevent future coronary events, such as heart attacks, in postmenopausal women who already had documented heart disease. In this study, the risk of coronary events increased during the first year on HRT but then leveled off: No further increase was seen. This result mirrors the pattern seen in the first 5 years of the WHI study, although in that study another significant increase in coronary heart disease was seen in the fifth year of HRT.

The fifth-year spike in the relative rate of heart disease in the WHI study, however, might not reflect a true increase in disease risk. In the first year, the incidence of coronary heart disease is 87% greater in subjects on HRT than in those receiving a placebo (see table). The rates of disease look fairly comparable in the second, third, fourth, and sixth years of the study; a substantial difference is seen in year five. But this apparent HRT-related danger arises from the anomalous decrease of events in the placebo group, not an increase in the incidence of heart disease in the HRT group. Yet this result, along with the increased risk of breast cancer that had been predicted as a possible adverse effect of HRT, led to the termination of the study.

[table legend] Relative occurrence rates of coronary heart disease, in episodes per 10,000 subject-years, of coronary heart disease.

Earlier observational studies noted that the rate of breast cancer increased by 30% after 4 years of HRT. In the WHI study, one-quarter of the participants had prior exposure to HRT, which means that they had been exposed to larger amounts of hormones than those women who were taking HRT for the first time. When the 25% with prior exposure were removed from the analysis, no increased risk of breast cancer was seen. So it appears that the WHI confirmed the observation that several years' exposure is necessary to increase the risk of breast cancer. As an aside, the risk of breast cancer is four times higher in women receiving combined HRT than in those taking estrogen alone. In the WHI study, women on Premarin alone showed a decreased risk of breast cancer, suggesting that the progestin, more than the estrogen, increases the risk of developing breast cancer.

WHI demonstrated that initiating HRT in older postmenopausal women can harm them--increasing their risk for breast cancer and stroke. It did not, however, answer the question of whether HRT could benefit women if initiated at the beginning of menopause. This window-of-prevention question will, in all likelihood, not be answered because volunteers and funding for future HRT studies are disappearing, a potentially regrettable consequence of the WHI findings.

Similar concerns can be raised regarding the other studies that were associated with WHI. For example, the Women’s Health Initiative Memory Study (WHIMS) showed a twofold increase in the risk of dementia in women taking HRT. Again, these findings were paradoxical. Many laboratory studies show that estrogen can protect neurons from toxins or other experimental assaults, yet the WHIMS results seem to suggest otherwise. A possible explanation for these discrepant observations was provided in the Cache County study of dementia, the results of which suggested that HRT could protect women against dementia, but only if the subjects began taking it around the onset of menopause; no protective effect was seen in those who started HRT within 10 years of the development of dementia. Because these cognitive disorders occur in older women, say ages 65 or 70, those subjects who had been taking HRT for less than 10 years before the onset of dementia most likely started HRT substantially after menopause set in. The 4532 women in WHIMS were all over 65, again raising the window-of-prevention question. Although one can reasonably conclude that there is no justification for putting an older woman on HRT if the intent is to try to forestall or prevent dementia, we still have no idea what role estrogen--or the loss of estrogen during menopause--plays in the development of dementia.

As for quality of life, the WHI trial claims to provide an answer, but it's unclear whether the study posed the optimal question. Jennifer Hays and colleagues from the WHI demonstrated that women who are not experiencing menopausal symptoms do not report an improved quality of life when taking Premarin. (Women who were symptomatic were discouraged from entering the study.) But it might be unreasonable to expect any treatment to improve the quality of life in an asymptomatic individual. Instead we should be asking whether women plagued by hot flashes and depression experience symptom reduction and improved quality of life from HRT.

Hays and her colleagues further obscured the answer to this question by minimizing their study's positive findings. Their results showed that women taking Prempro--estrogen plus progestin--described a significant improvement in their physical functioning, amount of pain, and sleep disturbance. But the authors describe these changes as clinically meaningless because the improvements seen in the group as a whole were small, and with so many people included in the study, the changes could have occurred by chance. Nonetheless, given that the overwhelming majority of subjects were asymptomatic and that HRT did not worsen their symptoms, these improvements should not be so easily brushed aside. Furthermore, the consistently positive effects on physical functioning, pain, and sleep at the 3-year analysis suggests that the benefits are real.

The importance of distinguishing between symptomatic and asymptomatic women is further supported by reports of the antidepressant effect of estradiol in perimenopausal women with major and minor depression. We observed that 3 weeks of 50 µg/day of estrogen, but not placebo, was a very effective antidepressant in women who developed depression just prior to menopause, findings that have subsequently been replicated by others. Consistent with the window of prevention concept, estradiol does not appear to be an effective antidepressant in women whose depression sets in years after menopause.

Finally, it appears that the results obtained with the type of HRT used in the WHI study, Premarin plus the progestin Provera, might not apply to other forms of HRT. Studies in animals suggest that Provera, unlike natural progesterone, might be particularly likely to produce adverse effects on the heart, brain, and breast. In other words, although synthetic progestins and natural progesterone both prevent excessive buildup of the uterine lining by estrogen, their effects may dramatically differ in other organs. A rose may be a rose, but a progestin is not a progesterone.

The purpose in raising these questions is not to suggest that the WHI study is without value. It was formerly all too easy to conclude, as many did, that estradiol is a panacea and that prophylactic HRT should be uniformly prescribed to decrease the burden of heart disease and osteoporosis. Such a position is clearly no longer tenable. HRT not only appears to have little prophylactic value in postmenopausal women, but it appears to be associated with some degree of increased risk, particularly of stroke and blood clots. However, we believe that it would be a mistake to conclude, on the basis of the available data, that HRT--and in particular, estrogen replacement alone--cannot protect against the development of cardiovascular disease. How a woman reacts to a drug or hormone depends on her health status when she begins treatment; consequently, therapies that have beneficial or protective effects in the absence of disease could have destructive effects in its presence. Although it is absolutely appropriate, therefore, to recommend against the uniform prescription of HRT for postmenopausal women, we could be throwing the baby out with the bath water if we don't attempt to better define the effects and potential benefits of HRT administered during the window of time when the physiological changes that HRT addresses occur: in the year or so surrounding the last menses.

This essay, then, is less an apologist's defense of HRT than it is an indictment of a one-size-fits-all approach to medicine. Physiology and pharmacotherapy are highly context-dependent. We neglect our obligations to our patients and the public if we fail to define those contextual factors and clinical predictors that will enable us to extract all possible therapeutic benefit from available treatments for the individuals we are trying to treat. The WHI study should be a significant chapter, but not the last chapter, in our efforts to understand the role that hormones and HRT play in health and therapeutics.

David R. Rubinow, M.D., and Peter J. Schmidt, M.D., are psychiatrists in the Behavioral Endocrinology Branch of the National Institute of Mental Health, NIH, in Bethesda, Maryland.

Additional Reading:

1. Writing Group for the Women's Health Initiative Investigators, JAMA 288, 321-333 (2002).

2. The Women's Health Initiative Steering Committee, JAMA 291, 1701-1712 (2004).

3. T. S. Mikkola and T. B. Clarkson, Cardiovasc. Res. 53, 605-619 (2002).

4. F. Grodstein, T. B. Clarkson, J. E. Manson, N. Engl. J. Med. 348, 645-650 (2003).

5. S. Hulley et al., JAMA 280, 605-613 (1998).

6. D. Grady et al., JAMA 288, 49-57 (2002).

7. S. Wassertheil-Smoller et al., JAMA 289, 2673-2684 (2003).

8. S. A. Shumaker et al., JAMA 289, 2651-2662 (2003).

9. P. P. Zandi et al., JAMA 288, 2123-2129 (2002).

10. J. Hays et al., N. Engl. J. Med. 348, 1839-1854 (2003).

11. P. J. Schmidt et al., Am. J. Obstet. Gynecol. 183, 414-420 (2000).