Hormone Studies: What Went Wrong?

By GINA KOLATA

The New York Times

22 April 2003

For nearly nine months, doctors and researchers have been struggling with an intractable problem: how could two large high-quality studies come to diametrically different conclusions about menopause, hormone therapy and heart disease?

The question arose in July, when scientists saw data from a large federal study called the Women’s Health Initiative, which was ended early when it became clear that a widely used hormone-replacement drug, Prempro, had risks, including heart attacks, that exceeded its benefits.

That finding directly contradicted previous studies showing that the hormones reduced heart disease risk — in particular, the Nurses’ Health Study, a large research effort that has been going on for years.

The question is why.

To answer it, researchers are reviewing the data, scrutinizing the design of each study and examining other research that may help reconcile the disparate findings. But so far, as they noted at a recent symposium at the Harvard School of Public Health, they have had no luck. As one explanation after another fails to hold up, the mystery deepens.

The effort is crucial because the contradiction has vast implications, not just for women’s health but for medical research in general.

This does not mean that doctors are changing their new advice on hormone therapy. In the wake of the Women’s Health Initiative, even many who once advocated the drugs after menopause now say women should use them only to relieve unbearable symptoms, like severe hot flashes, and should take them for the shortest possible time.

Even if it turns out that the drugs do protect some women against heart disease — that the Women’s Health Initiative did not tell the whole story — there are better ways of doing it, medical experts say, than taking drugs that also confer other risks.

But as their efforts to reconcile the two studies continue to fall short, the experts continue to worry.

“If there are such discrepancies, how many other medical facts do we believe based on one type of study or another type of study because we don’t have the luxury of having both?” asked Dr. Isaac Schiff, the chairman of obstetrics and gynecology at Massachusetts General Hospital in Boston.

The Nurses’ Health Study was what scientists call observational. Tens of thousands of nurses reported on their hormone use and their health. Researchers found that hormone users had 30 percent fewer heart attacks.

By contrast, the Women’s Health Initiative was the sort of study that is considered a gold standard — thousands of women were randomly assigned to take the most widely used hormone preparation, Prempro, made by Wyeth, or a placebo, and their health was monitored. Women who took hormones had 40 percent more heart attacks.

Ordinarily, a randomized placebo-controlled study like this trumps an observational study; that is what happened here. Within weeks, many doctors who had recommended hormone therapy changed their advice. Prempro sales plummeted.

But that is not the end of the story, for the Nurses’ Health Study is no ordinary observational study. Its participants are nurses, presumably better able than most people to report reliably on the drugs they take, exercise habits, diets, and so on. So its conclusions are hard to dismiss and, in fact, they have often helped shape public health policy on a number of issues.

Also, its findings were not an anomaly. Other similar studies of hormone therapy found the same protection from heart attacks among women who took the drugs. And it and the other studies agreed with the Women’s Health Initiative on other risks and benefits of the hormones: a slight increase in the risk of breast cancer, blood clots and strokes and a slight decrease in the risk of broken bones and colon cancer.

On the other hand, the Women’s Health Initiative results were in accord with similar studies of women who already had heart disease. Those studies also found that hormone therapy led to more heart attacks, not fewer.

As soon as he saw the results from the Women’s Health Initiative, said Dr. Barry R. Bloom, dean of the Harvard School of Public Health, “I knew that public health had a big problem.”

“The dilemma posed by this issue is very profound and quite important,” Dr. Bloom said. “Is there an intrinsic flaw in either of these approaches?”

Now researchers are asking which type of study is correct — or, perhaps, whether both could be correct, or both wrong.

Dr. Francine Grodstein, an epidemiologist at the Harvard School of Public Health, says it is quite possible that both are correct. The different results may hinge on the differences between the women who joined the studies.

The nurses who took hormones did so deliberately, and most continued with them for years. That is not typical behavior: most women who received prescriptions for hormones took them briefly or not at all, studies have found.

Women who stay with a prescription regimen may be unusual, Dr. Grodstein said. Perhaps they generally care more about their health, and that is the reason they have fewer heart problems. Or perhaps there was some other physical characteristic, as yet unidentified, that made estrogen protect them from heart disease.

Or maybe it was the participants in the Women’s Health Initiative who were atypical, Dr. Grodstein said. Most women would not volunteer to be randomly assigned to take hormones or a placebo for years on end. Is there something about the type of women who join such a study that indicates an increased risk of heart disease if they take hormones?

Maybe it is their weight: they were heavier, on average, than those in the nurses’ study. Since body fat produces its own estrogen, the added hormones they took might have pushed them over a threshold where heart attack risk would emerge. Moreover, two-thirds were over 60 when they joined the study and started taking hormones. In contrast, most women in the nurses’ study and ones like it started taking hormones at menopause.

But those explanations do not appear to be enough, said Dr. JoAnn Manson, a researcher at Brigham and Women’s Hospital in Boston, who was an investigator in the two studies. Women who were 50 to 59 when they joined the Women’s Health Initiative had the same increased risk of heart disease with hormone therapy as the older women in the study. Thinner women in that study, whose body fat was comparable to those in the nurses’ study, also were not protected from heart attacks when they took hormones.

Another possibility is that the nurses’ study and others like it miscounted hormone therapy users. The nurses’ study assesses drug use every two years. A woman could start taking hormones, have a heart attack in the next two years, and stop taking them immediately. Later, when asked if she takes hormones, she says, correctly, that she does not. Her heart attack would be counted as one that occurred in a nonuser.

But Dr. Grodstein added that this sort of error, by itself, was not enough to explain the discrepancy.

Some suggested that it was the particular drug, Prempro, studied in the Women’s Health Initiative that elicited heart attacks. The nurses took a variety of hormone preparations, not just Prempro.

But Dr. Jacques Rossouw, an administrator with the Women’s Health Initiative, said that several other randomized controlled trials of women who already had heart disease found that other hormone preparations also increased heart attack risk, an indiction that they may be no safer than Prempro.

Dr. Rossouw said all possibilities remained open: the nurses’ study and ones like it could be right and the Women’s Health Initiative could be wrong, or vice versa.

If each is right, he said, it may be because the women in the two types of studies are different in a way that researchers have not yet figured out. But if so, it means that hormones protect some women but are too risky for others. Then again, Dr. Rossouw said, maybe each study is wrong. Maybe estrogen, in pills, is not the chemical to focus on.

What is missing, he said, is an explanation for sex differences in heart disease. Throughout life, women have less heart disease than men, and the difference is especially pronounced at younger ages. He said it was possible that estradiol, the estrogen women naturally produce and that goes directly to their bloodstream is protective. The estrogen in pills, metabolized by the liver before entering the blood, may be harmful. Neither the nurses’ study nor the Women’s Health Initiative addressed that question, Dr. Rossouw said.

To answer it, he said, “we would have to do a different kind of trial.” It would maintain the estradiol levels of youth, which start to drop at about 45. The women would be younger, and the estradiol, which cannot be taken in a pill, would be in a patch or nasal spray.

One thing is clear: scientists cannot ignore the question.

Many women need hormone therapy Dr. Rossouw said. “We can’t walk away from it,” he said. “Women will continue to take it. A third of women get moderate to severe symptoms of menopause, and there really is no other option.”

In the meantime, he said, “the No. 1 research priority” is to discover why the women in the health initiative had more heart attacks. “What’s the mechanism?” Dr. Rossouw asked. “If we can find it, maybe we can find drugs that don’t trigger it.

“We’ve got to go back to the drawing board.”