The headlines should have swept the nation and the world – ESTROGEN REPLACEMENT THERAPY LOWERS BREAST CANCER INCIDENCE WHILE REDUCING MORTALITY AS WELL. But you probably never heard a word about it. And if you did, you probably heard the distorted version that has tricked and befuddled many, ever since the Women’s Health Initiative began over 20 years ago – HORMONE REPLACEMENT THERAPY INCREASES BREAST CANCER RISK. What gives?
The media coverage from the Women’s Health Initiative (WHI) has been shaky for many years, not to mention widespread misunderstanding by clinicians, which then translates to a confused public. This is not helped by local newspapers where the story from the newswire services (Associated Press et al) might report the story correctly, but since the lead can be buried at the tail end of the press release, it often gets the axe (as happened in our ever-shrinking Oklahoma City newspaper), and doesn’t appear in print at all. A devoted search online will eventually lead to the facts, but filtering out the garbage can be very difficult.
Let me see if I can resurrect the mind-blowing results of the WHI that have been inexplicably ignored for 20 years.
The Women’s Health Initiative (WHI) was launched by the NIH in 1993 addressing several health issues with proposed interventions for postmenopausal women – heart disease, breast cancer, colorectal cancer and osteoporosis/bone fractures. Of several high-quality WHI clinical trials, we are interested here only in the impact of hormone replacement therapy on breast cancer risk and mortality. And here’s where the confusion begins – Estrogen plus Progesterone had the opposite effect to using Estrogen alone.
Yet, media coverage (and downstream public confusion) regularly lumps the two trials with very different outcomes into one all-encompassing (and incorrect) statement: “The WHI showed us that hormone replacement therapy increases breast cancer risk.” The error is generated by the phrase “hormone replacement therapy,” which covers different concoctions, far too ambiguous to use in this controversy.
I’ve even heard experts at the podium boast about the WHI as proof that “evidence-based medicine” dispels gross misconceptions, e.g., “We once thought that hormone replacement therapy was safe with regard to breast cancer risk, but the WHI showed us the dangers, and now that women have decreased their hormone use, breast cancer incidence is actually declining.” By the end of this article, you should be able to see the error in that seemingly benign statement.
Twenty years ago, I wrote the first lay book on breast cancer risk assessment. Well, more accurately, it was almost the first. Dr. Patricia Kelly’s book (Assess Your True Risk of Breast Cancer) was released one month before mine, and we both essentially wrote the same thing, independently. Yes, bookstores were loaded with tomes covering various risk factors for breast cancer, but the distinction is this – no authors of lay literature were combining multiple risks into a workable number from which patients could make rational decisions. This would later become known as “breast cancer risk assessment.” The Gail Model had been developed, but was largely unknown at the time except for its employment in the NSABP P-01 trial on tamoxifen prevention. Other models had been proposed in the epidemiology literature, but had not been recognized yet at the clinical level. Anyway, in that book, I distilled what was known at the time about hormone replacement therapy (HRT), and it was clear that there was a difference between estrogen alone (in women who had undergone prior hysterectomy) and estrogen plus progesterone (in women with an intact uterus).
Note: the progesterone is added primarily to protect the lining of the uterus from the non-stop estrogen effect, which can increase the risk of uterine cancer. Sadly, many patients back then were told that E+P would be good for reducing breast cancer as well (“just like the uterus”), but evidence was already available 20 years ago that indicated otherwise.
Although the available literature at that time was based on observational studies without prospective, randomized trials, there was a consistent feature – E+P was worse than E alone. In fact, I made the bold statement (bold in 2000, that is) that if you made the distinction between these 2 very different HRT approaches, future studies might prove that estrogen alone will carry no risk at all. This was nearly heresy. But I wasn’t relying only on available clinical literature at that time. I had reviewed the basic science literature as well. Histologic changes in breast tissue had been documented during different phases of the menstrual cycle, as well as various postmenopausal hormone preparations. Oddly, estrogen alone had very little impact on mitotic figures in the breast (or on other proliferative markers), which indirectly might translate to no increased breast cancer risk.
As it turned out, my prediction of E-alone as being “neutral” was not revolutionary enough. The WHI showed a lower risk of breast cancer in the E-alone limb from the git-go. And this finding was roundly ignored, with the entire focus placed on the E+P findings that prompted stoppage of this E+P limb of the study due to the increased breast cancer risk (the increased risk being a predetermined checkpoint that was actually a rather modest degree of risk, but nonetheless real). The poor E-alone findings of benefit got dragged down to the depths of obscurity even though the results were opposite that of E+P. In the end, it was all lumped together as HRT, in general, “increases breast cancer risk unacceptably.”
It is critical to understand that these were two separate trials within the WHI umbrella – 1) Estrogen (E-alone) vs. placebo and 2) Estrogen plus Progesterone (E+P) vs. placebo. Both were prospective, randomized trials, which have more power in defining truth than observational studies, even when those lesser studies are combined (selection bias in observational studies can only be limited through prospective randomization). 16,608 women participated in the E+P trial, while 10,739 participated in the E-alone trial. In the E-alone trial, women had previously undergone hysterectomy, thus eliminating the risk of uterine cancer in those women who take estrogen alone.
As noted above, when the initial report was released in July 2002, it prompted early termination of the E+P group as risks were outweighing benefits, led by the predesignated breast cancer risk. Even then, however, as “HRT” was widely blackballed, the results in the E-alone limb were surprising – there was a LOWER incidence of breast cancer. Experts dismissed the finding as it failed to reach statistical significance (though it was close), and the anti-E+P juggernaut continued, incorrectly absorbing results of both trials in its wake as a single finding in a single study (through no fault of the WHI, but through commentators on the trial).
But then something unexpected happened with longer follow-up – the E-alone women had fewer and fewer breast cancers emerge until that group did, in fact, reach statistical significance. AND YOU DIDN’T HEAR A WORD ABOUT IT! Why? I’m not sure, except that it didn’t make biologic sense (to most). The E+P findings made sense, and the risk was readily accepted. But the E-alone lowering of risk had no biologic explanation.
Improbably, E-alone appeared to be preventing breast cancer. Pundits continued to point out the wonder and beauty of evidence-based medicine in reducing breast cancer incidence through widespread stoppage of E+P, even though these same experts and journalists were simultaneously ignoring the shocking evidence in the E-alone trial. “Well, just give it time, and you’ll see that E-alone will eventually raise risk.” Or, so it was said by the few who actually had noticed the disparate findings between the two limbs.
In 2004, the E-alone trial was stopped as had been done previously for the E+P trial. But it was not because of breast cancer risk, which had landed on the benefit side of the scale. It was because of a small increase in strokes in those taking Estrogen alone over placebo. Without digressing too much, this stoppage was inconsistent with the SERM risk reduction trials where both tamoxifen and raloxifene demonstrated an increased risk of thrombotic events, but the chance of a breast cancer being prevented greatly exceeded this risk, allowing FDA approval for breast cancer risk reduction with SERMs. But the WHI had a different philosophy, largely because so many more women would be taking Estrogen if endorsed. Even though the risk of stroke was very small in comparison to overall benefits, they did not think a preventive medication should carry any measurable risk. For purposes here, the E-alone trial, unlike E+P, was not stopped due to the breast cancer risk.
But in December 2019, at the San Antonio Breast Cancer Symposium, Rowan Chlebowski, MD, PhD, stunned the audience with the 20-year follow-up for the TWO hormonal therapy studies of the WHI. The E+P results were similar to earlier reports – a 29% higher incidence of breast cancer, statistically significant, and a 45% increased risk of breast cancer death. It is here that stoppage of E+P (at least the combination used in the WHI – Prempro™) has likely spared many women the diagnosis of breast cancer. But in fact, this was nothing new. What was new about the E+P follow-up was the fact that the risk persisted even after stopping the hormones. In the first reports, this risk seemed to disappear, then later reports indicated some risk persisted, and now the long-term follow-up suggests the E+P risk to be more persistent than we thought after stopping therapy, perhaps lasting a lifetime.
But the “shock” came with the other study — the long-term follow-up in women who took Estrogen alone vs. placebo. The lowered incidence of breast cancer on estrogen was maintained and now clearly significant from the statistical standpoint – a 23% reduction in breast cancer incidence (remembering that E+P increased risk 29%). But much to the surprise of even Dr. Chlebowski, there was a statistically significant reduction in breast cancer mortality of 44% with the use of estrogen alone.
As Dr. Chlebowski pointed out, the FDA has approved two drugs to prevent breast cancer – tamoxifen and raloxifene – and neither of these agents has been shown to lower breast cancer mortality. Yes, both agents lower breast cancer incidence (slightly more powerful than the estrogen data above), but neither drug used in the preventive setting has yet been show to lower morality due to breast cancer. Only Estrogen fits that bill. Go figure.
Another oddity is that the reduced risk occurred primarily in estrogen-positive/progesterone negative tumors. How does taking estrogen reduce the risk of an ER-positive tumor? Because the story is much more complicated than what one chooses to take as replacement therapy, and the changes in ER status of individual cells likely take place largely independent of what hormone one takes above and beyond the natural milieu. Patients routinely state things like, “My mother’s breast cancer was estrogen-fed, so I’m not too worried about my family history since I don’t take estrogen.” This is faulty reasoning, and such a patient should acknowledge the family history independent of whether the tumor is “estrogen-fed” or not.
And another point: all the percentages above are referring to RELATIVE RISK, which can be very misleading. Relative risks are always more impressive than ABSOLUTE RISKS. An increase from one cancer to two cancers in a study would be a “100% increase” in relative terms. A decrease from two cancers to one, on the other hand, is a “50% decrease in risk” in relative terms. Relative risks are calculated without knowing the total number of patients in a study. Absolute risks include that number, offering far greater perspective in making therapeutic choices (and much lower numbers that aren’t as exciting).
For example, in the E+P study, there were 1,003 cancers overall, so if 438 cancers occurred in the placebo group, then a “29% increase in risk” would generate 565 cancers in the group that took the combination of E+P, or an excess of 127 cancers attributable to the use of E+P. But there were about 8,000 women who took E+P and did NOT develop breast cancer, so this excess of 127 cancers translates to an absolute risk of 1.6% that E+P will generate a breast cancer above and beyond baseline risk. This absolute risk might be acceptable to an individual trying to make a decision about HRT using E+P, but on the other hand, it is quite unacceptable to public health experts who translate that tiny number to the millions of women who were taking E+P at the time, prompting discontinuation of the study.
Turning to the E-alone study, there were 520 cancers in the long-term follow-up, so if 294 cancers occurred in the placebo group, then a “23% reduction” would leave us with 226 cancers in the group that took estrogen. But there were over 5,000 women who took estrogen in this study who did NOT develop breast cancer. So, the benefit was restricted to 68 women (294-226=68) out of 5,000, around 1.4%. Which is more impressive – a 23% relative reduction, or a 1.4% absolute risk reduction? Both are true. “Relative” wins the popularity contest every time, but it’s the 1.4% absolute benefit that gives you better perspective.
(Note: It was this sort of exercise – distinguishing relative risks from absolute risks – that Dr. Patricia Kelly and I introduced to the lay public, and many professionals, in our books published in year 2000. It’s hard to believe now, but very few studies made the distinction at the time. And while both books could make the reader math-weary, the bulk of information included was brand new to both the lay public and health care providers.)
One month has passed since the WHI announcement in San Antonio, and very little can be found even online about this landmark study. And when discussed more precisely in terms like “combined hormone therapy increases breast cancer risk and mortality, an opposite effect than single agent therapy,” the impact of this study can still be missed. That is, estrogen-alone replacement therapy might actually lower breast cancer risk and mortality. Few readers of current articles could explain exactly what “combined therapy” or “single agent therapy” is. Only rarely will you see a distinction made, using the culprit’s true name – progesterone (in combination with estrogen).
So, does the scientifically “pure” WHI study, with over 10,000 women in the E-alone study, trump the 100,000 women in the U.K. meta-analysis that combined 58 lower-quality studies where estrogen alone had a modest increased risk for breast cancer? Not all 58 studies showed increased risk, of course, but the combined meta-analysis revealed an increased incidence with estrogen, regardless of the exact preparation or whether oral vs. transdermal (only vaginal estrogen cream escaped implication). Hard to say.
Observational studies include a variety of built-in biases, such as “selection bias,” a phenomenon difficult to erase even with meta-analyses. For example, women who opt for estrogen replacement seem to be more compliant with routine screening mammography, and if so, more low-grade cancers will be identified – not because of a negative impact from estrogen but due to the “length bias” potential (call it overdiagnosis, if you must) of screening. Meanwhile, those not taking estrogen might be less likely to get mammography whereupon fewer cancers will be found.
As for me and my patients who ask about estrogen-alone replacement, I’ve been telling women about the WHI ever since their first announcement 18 years ago, using words like this: “The most scientifically pure study ever performed indicates estrogen alone is safe with regard to breast cancer risk and might even be protective against breast cancer,” often adding: “new studies might show different results, or that non-stop use for decades might negate the benefit seen with short-term use, but for now, it appears unlikely that you could do yourself harm with estrogen alone.” And if more information is requested, I go into absolute risk and benefit, as distinct from relative risks and benefits, especially if we’re talking about using E+P.
Although the story is still being written, my question is more than “Does estrogen alone actually prevent breast cancer deaths?” The more immediate question is: “Why is no one talking about it?”