In this post, I’ll expand on the claim I made in “Hormone Replacement Therapy is Much Better and Much Safer Than You Think” that there is no solid evidence that smoothly transitioning into hormone replacement therapy (HRT) or estrogen replacement therapy (ERT) at the first signs of menopause causes breast cancer.

I strongly encourage you to read the book Estrogen Matters, by Avrum Bluming and Carol Tavris, to get the full argument. All the indented quotations in this post come from that book. Here, I’ll just give a taste.

First, to define the terms ERT and HRT, they write:

… women who have had hysterectomies and who subsequently start hormone therapy get estrogen alone (ERT), while women who have not undergone hysterectomies and who start hormone therapy receive estrogen plus progesterone (HRT).

There is a lot of evidence for a lack of correlation between HRT or ERT and breast cancer as HRT and ERT are normally used. Here are some of the points Avrum Bluming and Carol Tavris lay out:

• A 1986 study led by epidemiologist Louise Brinton at the National Cancer Institute found no statistically significant increased risk of breast cancer among women on Premarin, even among those who had been taking it for more than twenty years.

• A 1988 meta-analysis of twenty-two studies by Bruce Armstrong at the Research Unit in Epidemiology and Preventive Medicine of the University of Western Australia found no statistical association between ERT and breast cancer.

• A 1991 study led by epidemiologist Julie Palmer at Boston University School of Medicine found no increased risk of breast cancer among Premarin users even after fifteen years of use.

• A 1991 analysis of twenty-eight studies by biostatistician William Dupont and pathologist David L. Page at the Vanderbilt University School of Medicine found no association between ERT and breast cancer.

• In 1992, the first randomized, double-blind, placebo-controlled trial on this subject was published. Twenty-two years earlier, obstetrician-gynecologist and medical researcher Lila Nachtigall and her colleagues at New York University Langone Medical Center had randomly assigned 168 postmenopausal women who were continuously hospitalized in a mental institution to receive either HRT or placebo. After more than two decades, 11.5 percent of the women taking the placebo had developed breast cancer—but none of the women on HRT had.

• Because women who undergo biopsies for benign breast disease have a slight increased risk of breast cancer, researchers followed 3,303 women who had benign breast biopsies performed at Vanderbilt University between 1958 and 1960. The median duration of the follow-up was seventeen years. In this study, published in 1989, women who were given estrogen following the biopsy—even those who had a family history of breast cancer—did not subsequently have an increased risk of breast cancer themselves.

Of course there were a few contradictory studies—there always are in medicine—but by the year 2000, major journals, research institutions, and leading oncologists were coming to the consensus that estrogen did not increase the risk of breast cancer.

…

According to a 1995 study by epidemiologist Janet Stanford of the University of Washington, “The use of estrogen with progestin (HRT) does not appear to be associated with an increased risk of breast cancer.… Compared with nonusers of menopausal hormones, those who used estrogen-progestin HRT for eight or more years had, if anything, a reduced risk of breast cancer.”

A 1995 article in the New England Journal of Medicine reported the first wave of results from the Nurses’ Health Study, which involved 121,700 female registered nurses who were followed from 1976 through 1992. The women who had used HRT at any point, even those who had been taking it for more than ten years, had no increased risk of breast cancer compared to women who never took HRT.

Some studies showed negative correlations with breast cancer:

In 1997, epidemiologist Thomas Sellers at the Moffitt Cancer Center of the University of Minnesota studied a random sample of 41,837 female Iowa residents between fifty-five and sixty-nine years of age to determine whether HRT was associated with an increased risk of breast cancer in women with a family history of breast cancer. It was not.

In 2006, biostatistician Masahiro Takeuchi, at the National Cancer Center Hospital in Tokyo, studied nine thousand Japanese women and found that those who were on HRT were less likely to develop breast cancer than never-users.

What got the notion that HRT and ERT cause breast cancer into a large number of heads—likely including the head of your primary care physician—was the Women’s Health Initiative and how its results were spun. (“How its results were spun” is not an exaggeration. On that, at this point, let me refer you to the book. Here I’ll just focus on trying to interpret the science appropriately.)

The Women’s Health Initiative (WHI) had the great strength that it was a large randomized control trial:

The WHI was the largest prospective study in which women were randomized to take either hormones or a placebo and then followed over time.

The reason a randomized controlled trial (RCT) is important is that it could be that women who got HRT or ERT were following healthier practices in other ways, that were not captured by the data collected. Note that if this what confounded cross-sectional analyses and hid a bad effect of HRT or ERT in an RCT, it would mean that, at worst, any extra breast cancer risk from ERT or HRT could be fully counteracted by lifestyle changes likely to be reasonable since they would correspond to lifestyle differences actually seen across different people.

Being a randomized controlled trial made the Women’s Health Initiative expensive and gave it a lot of prestige, but it had other weaknesses I’ll discuss below. One thing to keep in mind is that even a relatively large randomized control trial would have low precision in its statistical estimates until a long enough time had passed for many women to have gotten breast cancer in the normal course of things. But, in what I regard as a mistake, the Women’s Health Initiative was stopped early on.

The initial results of the WHI were equivocal, within the bounds of what could be a statistical fluke from the particular women who were in the study and their idiosyncrasies:

Let’s start with the claim about breast cancer. The WHI investigators reported that women who were randomly assigned to take estrogen on its own had had no increased risk of breast cancer. Those who still had a uterus and were assigned to take the combination of estrogen and progestin had a small increased risk of breast cancer (1.26) when compared with women who were randomly assigned to a placebo. That number, 1.26, would mean a 26 percent increase in risk. What few noticed was this sentence: “The 26 percent increase in breast cancer incidence among the HRT group compared with the placebo group almost reached nominal statistical significance.” Almost means it did not reach statistical significance, and that means it could have been a spurious association.

Importantly, as time went on, the suggestive evidence for some additional breast cancer risk disappeared:

In 2006, in another update of this same cohort of women, the WHI reported that they found no increased risk of breast cancer among those same women randomized to combined estrogen-progestin treatment. The alleged increased risk—the one worth stopping the study for—had completely vanished.

Because the trial was stopped prematurely, we can’t know what the effects of HRT or ERT for longer would have been, but the lack of aftereffects certainly should be reassuring.

It is a stretch to read the WHI as showing any convincing evidence of higher breast cancer risk at all. But to the extent that you do read it that way, the weirdness of the sample means it wouldn’t say much about the way ERT and HRT are normally done. The biggest problem was that older women well past menopause were a dominant part of the sample, so the WHI can’t say much about the effects of going smoothly into ERT or HRT at the first sign of menopause. Another problem is that it is mostly speaking to breast cancer risk of women who have other problems, and so is less informative about any potential dangers of HRT or ERT for a woman who is healthy to begin with:

The WHI was heralded as being truly representative of women during and after menopause, and the WHI investigators repeatedly stated that all of the women they recruited were healthy at the outset of the study, but neither assertion was true. Fully 35 percent of the women were considerably overweight, and another 34 percent were obese; nearly 36 percent were being treated for high blood pressure; nearly half were either current or past cigarette smokers. Moreover, the median age of participants was sixty-three, long past the onset of menopause. Therefore, there is no credible reason for generalizing from the results of this study to the entire population of postmenopausal women—even though that was precisely what this randomized controlled study was supposed to do.

Above, I wrote “Note that if this what confounded cross-sectional analyses and hid a bad effect of HRT or ERT in an RCT, it would mean that, at worst, any extra breast cancer risk from ERT or HRT could be fully counteracted by lifestyle changes likely to be reasonable since they would correspond to lifestyle differences actually seen across different people.” To spell that out more, let me select from Avrum and Carol’s long list of risk ratios some of the more interesting risk ratios that given context to the preliminary (and possibly statistically fluke) result of a 1.26 risk ratio for breast cancer of HRT in the WHI. I’ll make my selected items into bullets without any ellipses in between. But what is after each bullets is a quotation. Avoiding something with a relative risk factor of more than 1.26 or doing something with a relative risk factor of less than 1/1.26 = .79 would cancel out a 1.26 risk factor from HRT if that risk factor from HRT turned out to be real. Here are my selected items:

Risk Factors Reported to Be Associated with Breast Cancer

• Risk Factor: Dietary fiber intake Relative Risk 0.31

• Risk Factor: Significant weight gain from age 21 to present Relative Risk: 0.52

• Risk Factor: Garlic and onions 7 to 10 times a week Relative Risk: 0.52

• Risk Factor: Fish oil Relative Risk: 0.68

• Risk Factor: Aspirin Relative Risk: 0.80

• Risk Factor: Coffee consumption more than 5 cups a day Relative Risk: 0.80

• Risk Factor: Above average weight at the age of 12

• Risk Factor: Exposure to light at night Relative Risk:

• Risk Factor: Alcohol Relative Risk: 1.26

• Risk Factor: French fries (1 additional serving per week during preschool years) Relative Risk: 1.27

• Risk Factor: Night-shift work Relative Risk: 1.51

• Father at least 40 years old at patient’s birth (premenopausal breast cancer) Relative Risk: 1.90

• Risk Factor: Antibiotic use for more than 1,001 days Relative Risk: 2.07

• Risk Factor: Increased carbohydrate intake Relative Risk: 2.22

• Risk Factor: Calcium channel blocker for more than 10 years Relative Risk: 2.40

Outside the realm of breast cancer, the risk ratio for tobacco smoking as a risk factor for lung cancer is 26.07.

I find Estrogen Matters a very important and useful book. It is good to remember that Avrum Bluming is an MD, trained just as well as your primary care physician, and probably quite a bit smarter than your primary care physician. Of course, your primary care physician knows things about you in particular that a book doesn’t reflect. So if your primary care physician advises against ERT or HRT for you based on unusual things about you in particular, you should take that very seriously. But if your primary care physician tends to oppose HRT and ERT across the board, it is probably because they are misinformed. Don’t be afraid to argue with your primary care physician; you can say you might change your mind if they read the book and give you a good argument after that. Also, you might be surprised at how different the opinions of different physicians are. There are some dangers to doctor shopping—and certainly some extra costs given the usual managed care—but it might be worth it in this case.

In this post I emphasized the absence of a breast cancer cost to HRT and ERT. Make sure to read “Hormone Replacement Therapy is Much Better and Much Safer Than You Think” to see more about the benefits. And both of these posts are meant to motivate you to read the book before you make any decision about ERT or HRT for yourself. You need to read the book so that your primary care physician doesn’t railroad you.

I don’t mean to say that the book is totally without flaws. There is some overemphasis of conventional significance levels, when it probably makes more sense to be a Bayesian in this context. And while I tried to provide context, breast cancer is common enough that even with the quite high survival rates for women who get breast cancer, I can’t dismiss a risk ratio of 1.26 as small if it were real and applied to your situation, which it probably doesn’t. But many of the known benefits of HRT and ERT are quite large compared to this, even it were real and applied to your situation. Nevertheless the book is great!

For organized links to other posts on diet and health, see:

• Miles Kimball on Diet and Health: A Reader's Guide