Meir Stampfer

In 1998, however, this assumption was questioned by a report of the randomized Heart and Estrogen/Progestin Replacement Study (HERS) clinical trial, saying that postmenopausal hormones did not protect women with existing heart disease from subsequent heart attacks. More questions were raised in July 2002, when data from the federally funded Women’s Health Initiative (WHI) trial of hormone therapy, reported in the Journal of the American Medical Association, showed that daily estrogen and progestin therapy failed to reduce the risk of heart attacks. In fact, hormone therapy appeared to raise heart attack risk soon after initial use, slightly but significantly. Millions of women and their doctors were thrust into uncertainty.

"The dilemma posed by this issue is quite profound," said HSPH Dean Barry Bloom as he opened "Controversies in Postmenopausal Hormones and Heart Disease" on February 13 to a standing-room only audience in Snyder Auditorium.

A webcast is available at www.hsph.harvard.edu/video/.

"To whom does a woman turn for the best scientific evidence to make a major life decision about health?" asked Dean Bloom. "The answers are of terribly great importance."

Organized by Meir Stampfer and Francine Grodstein, both investigators of the NHS, the symposium included epidemiologists, physicians, one of the leading scientists of the WHI, and a basic scientist studying the cardiovascular effects of hormones in animals. Stampfer is chair of the Department of Epidemiology and Grodstein is assistant professor of epidemiology, both at HSPH.

At the end of the two-hour symposium, Stampfer described the exchange as the "most stimulating session I’ve attended in 20 years of research on estrogen."

Isaac Schiff of Massachusetts General Hospital, who specializes in treatment of menopausal symptoms, served as moderator. When the results of the WHI trials were published, he said, "The people who felt most betrayed were our patients. They had had a generation of information saying they should take estrogen to prevent heart disease, Alzheimer’s, and other diseases."

He said that despite the disappointing findings in heart disease prevention, hormone therapy "is the best agent if a person is symptomatic; she deserves to be treated."

For most women without a family history of heart disease, the biggest worry from hormone therapy is not heart attack, but breast cancer.

Grodstein led off the program by discussing an array of possible explanations for the apparently divergent findings. She was also lead author of a "Sounding Board" article on the topic in the New England Journal of Medicine, published on the same day as the symposium.

In the NHS, the health of more than 100,000 nurses has been studied for nearly 30 years. The researchers have used intermittent questionnaires to search for, among many issues, any difference in heart disease risk in women who chose hormone therapy. The WHI, by contrast, was a clinical trial in which postmenopausal volunteers were randomly assigned to begin hormone therapy or to refrain from it. They were closely followed for a number of years and the incidence of heart attacks and other events compared between the two groups.

Grodstein pointed out that despite their different study methods, the NHS and WHI results agreed on the relative risks to hormone users and non-users on all the conditions studied–with the puzzling exception of coronary heart disease. (Other conditions studied included invasive breast cancer, colorectal cancer, hip fracture, stroke, and pulmonary embolism.)

The NHS had found a 50 percent CHD risk reduction, while conversely, the WHI found that hormone therapy raised the risk by 29 percent. The similar findings for the non-CHD related outcomes (especially stroke) suggest that confounding is unlikely to explain the discrepancy, unless CHD risk might be uniquely susceptible to confounding.

Grodstein grouped the potential biases in the observational studies into three types. The subjects might have differed in health and lifestyle practices or in factors such as socioeconomic status. There might have been biological disparities among the women or in the hormone regimens. Or the methods used in observational studies versus clinical trials might have contributed to the conflicting outcomes.

In her survey of possible explanations, Grodstein named two as most likely:

First, if starting hormone therapy heightened cardiac risk right away, the WHI trial would spot resulting heart attacks more easily than the observational Nurses' Health Study, which only checked for hormone use and heart attacks every two years.

Second–and this possibility received substantial attention at the symposium–the women in the WHI trial were older and most started taking hormones 10 years or more after menopause. By contrast, the NHS nurses typically began right after menopause. Could the hormones exert different effects on heart disease depending on a woman’s age?

The question led directly to the hypothesis of the next speaker, Thomas Clarkson, professor at Wake Forest School of Medicine.

For many years, Clarkson has studied the effects of hormone therapy at different ages in female primates made surgically menopausal. He has compared this timeline with the evolution of coronary disease as it begins in young women, accelerates rapidly from ages 40 to 50 as estrogen production dips, and then results in inflammatory and calcification damage to arteries after age 55.

"Hormone therapy seems very effective at inhibiting the early process," said Clarkson, "but we have data saying that, in the later stages of coronary heart disease, it [hormone therapy] has an adverse effect."

When initial hormone treatment was delayed in the animals by an amount equivalent to six years after menopause in humans, he added, "There was no longer any benefit at all."

To Clarkson, these findings suggest a very plausible mechanism that could explain an apparent cardiovascular benefit in the younger NHS volunteers, while the same hormone regimen may have disrupted atherosclerotic deposits in the older WHI subjects, causing the formation of clots that triggered heart attacks.

In the final main talk, Jacques Rossouw, physician and lead project officer for the Women’s Health Initiative, argued that there were enough potential unresolved biases in the conflicting study results to explain three possible outcomes: that is, the NHS could have overestimated the benefit of hormone therapy in heart disease, the NHS and WHI could both be correct for their respective populations, or both could have missed the mark. The observational studies found that users of oral hormones appear to have lower risks of CHD, and this is the hypothesis that was tested in clinical trials (though the women in the clinical trials began hormone use at older ages than in the observational studies).

However, testing of a somewhat different hypothesis–that the gender difference in onset of CHD in healthy women is due to endogenous estradiol–will require a different kind of clinical trial of non-oral estradiol in younger, healthy, immediately post-menopausal women, said Rossouw.

He called for further studies, including studies to identify the mechanisms of risks of CHD in hormone trials, followed by short-term trials of various hormone combinations and delivery systems.

"Women will continue to use hormones for menopausal symptoms," he said. "We need hormones with a better cardiovascular profile, with no short-term risk, and hopefully with long-term benefit."

Participants in the panel discussion briefly commented on points in the talks or brought still other issues to the table.

JoAnn Manson, professor in the Department of Epidemiology at HSPH and co-investigator of both the NHS and WHI studies, said she was digging deeper into the data to determine whether any subgroups of the volunteers had some benefit from the hormone regimen, but so far "estrogen and progestin do not seem to be lowering coronary heart disease in any subgroups," she said.

Manson recommended further randomized trials including recently menopausal women using various forms of hormone therapy. However, she said, "I think the coronary heart disease question is somewhat of a moot point: if hormone therapy increases the risk of breast cancer in three to four years, and you need three to four years for it to help prevent heart disease, it’s still not going to be a viable option."

James Robins, professor of epidemiology at HSPH, noted that further statistical analysis of the Nurses' Health Study could address some of the potential "lifestyle" biases raised by critics. For example, some scientists have suggested that women who stayed on hormone therapy (adherent women) appeared to have lower risk because of that characteristic rather than because of the hormones. Using a simple diagram, Robins demonstrated how the potential bias of compliance could be factored out by analyzing the data as one would in a randomized clinical trial, instead of an observational study.

Along with epidemiology and clinical research, basic science has an important role in probing the hormone therapy riddle, said panelist Richard Karas of Tufts University.

"One of the take-home messages is that the effects of hormones on the cardiovascular system are very complex," he said.

For example, estrogen enhances the proliferation of endothelial cells but inhibits the proliferation of vascular smooth muscle cells, and these actions depend on complex molecular signaling pathways. New compounds such as the selective estrogen-receptor modulators, or SERMS, hold "enormous promise" for manipulating these signaling pathways more safely, said Karas.

The final panelist, Mary Beth Hamel, deputy editor of the New England Journal of Medicine, said that as a physician, she had to rely on clinical trial results in making recommendations to her patients. Regardless of the debate, she said, "I think it’s unlikely that we are going to use these drugs to prevent heart disease. The remaining question I’m interested in as a woman and as a doctor is how big are the risks–if they do exist–for short-term use for symptoms? My sense is that the risks are quite small, but they are real and should not to be discounted."

--Richard Saltus