Despite decades of scientific scrutiny, the mechanics of how different types of estrogen interact with cancer is not well understood. Some scientists say the sex hormone protects women from getting breast cancer; others say it fuels the growth of cancer cells.

Rachel Carson, who died of breast cancer two years after publishing “Silent Spring” in 1962, considered the synthetic estrogens found in drugs and cosmetics to be possible carcinogens. In the 1970s, epidemiological studies of cancer incidence rates from local cancer registries tethered an epidemic of 15,000 cases of uterine cancer to the long-term use of replacement estrogen. Sales of the drugs plummeted, and the epidemic abated.

Several years later, pharmaceutical companies reformulated the product, and sales bounced back.

Wyeth Pharmaceutical’s popular Prempro combined estrogen with the synthetic steroid hormone progestin. Wyeth advertised the drug, also known as E+P, as protective against uterine cancer and heart disease.

In 1991, the National Institutes of Health launched a $725 million project called the Women’s Health Initiative that sponsored, among other studies, a randomized clinical trial examining the effects of estrogen and progestin replacement therapy on 17,000 post-menopausal women.

In 2002, to great media fanfare, the Women’s Health Initiative abruptly terminated the trial. Researchers said Prempro had increased the risk of heart attacks and breast cancer, outweighing benefits like the alleviation of hot flashes.

Today the belief that hormone replacement therapy can cause breast cancer prompts large numbers of menopausal women to eschew the relief afforded by hormone therapies.

But they need not worry.


The ripples of a flawed trial

When the health initiative cancelled the trial, it announced that the cohort receiving E+P had incurred 26 percent more invasive breast cancers than had those taking a placebo.

The difference was couched in terms of relative risk, which, as we learned in the second installment of this series, can overstate the significance of small differences between cohorts. Absolute risk is a better measure of significance.

The absolute risk revealed by the E+P study was far less disturbing. The annual difference in invasive breast cancer risk between the cohort taking E+P and the cohort taking a placebo translated into eight cases per 10,000 women, or 8/100s of 1 percent, a difference small enough to be explained by chance or a flawed study design.

The initiative was widely criticized for packing the E+P trial with women who were older than 51, the normal age when menopause begins and women begin taking hormones to relieve symptoms.

This was done by design, so that participants, whose average age was 62, could not tell which cohort they were in by experiencing whether or not symptoms of menopause were being alleviated by the unmarked pills.

But it meant that the trial was not representative of women who take hormones to relieve active symptoms of menopause. It also meant that women who were not experiencing symptoms were ingesting hormones they did not need. Excess estrogen can increase breast density, which increases the risk of false positive diagnoses.

Critics also said the trial included many women who were slightly obese or who were former smokers, factors that increased their chances of being diagnosed with cancer.

And since the disease can take more than seven years to develop, some of the women diagnosed with breast cancer during the multi-year trial were likely to have had the disease before they were dosed with Prempro.

Though it is theoretically possible that E+P plays a role in causing or promoting the spread of breast cancer, the terminated study did not make that case.

And though it was stopped on the basis that the possible harms of hormone therapy outweighed its benefits, the health initiative later concluded that E+P has a protective effect for colon cancer. Another study the initiative sponsored found that postmenopausal women with prior hysterectomies who took estrogen alone had no increase in breast cancer incidence.

Nothing is cut and dried in epidemiology—especially when small absolute differences can be accounted for by study bias and random fluctuations in the data.

Nonetheless, the termination of the Prempro trial spurred media outlets across the world to declare that hormone replacement therapy causes cancer.

Over the next several years, about half of the 90 million Americans who had been prescribed hormone replacement drugs stopped using them. These women also reportedly curtailed menopausal-related doctor visits by 16 percent, lowering the rate of mammography referrals that had been obligatory before women could be prescribed hormone replacement treatments.

In the year following the study’s termination, the rates of screening mammography plunged as women stopped taking the feared drugs.
According to a national cancer registry, known as the Surveillance, Epidemiology, and End Results Program, or SEER, the incidence rate of newly diagnosed breast cancers dropped by 10 cases per 100,000 women, or 10/1,000s of 1 percent, between 2002 and 2003.

There was virtually no change in absolute terms.

Nonetheless, some scientists and the media framed the tiny change in breast cancer incidence as a drop of nearly 7 percent, the relative difference.

The fractionally falling incidence rates can be accounted for by chance or decreasing rates of breast cancer screening. According to federal data, the use of mammography dropped 4 percent from 2000 to 2004, with a precipitous dip in 2003; that rate gradually rose back up over the next decade.

Importantly, the screening rate for menopausal women, nationwide, dropped by 6.8 percent from 2000 to 2005. And the number of women getting diagnostic mammograms plummeted by a third: from 3 million in 2000 to 2 million in 2003.

A federal report on mammography largely attributed the massive drops to a sudden decrease in funding for screening.

In 2006, Peter Ravdin, a researcher at the MD Anderson Cancer Center in Houston, prematurely announced the results of a study on the 2003 incidence rate drop. The report, titled “The Decrease in Breast Cancer Incidence in 2003 in the United States,” was co-authored by the chief epidemiologist at the National Cancer Institute, Brenda Edwards, and others.

At a breast cancer symposium sponsored by GlaxoSmithKline, Dr. Ravdin portrayed the 7 percent decline in breast cancer incidence from 2002 to 2003 as astonishing. Without releasing the supporting data, he speculated that the drop was a result of millions of women suddenly abandoning E+P after the termination of the clinical trial.

Dr. Ravdin said “anecdotal” reports showed that E+P acted as a “fuel” for existing breast cancers; removing the fuel had instantly and dramatically reduced the number of cancers, causing them to regress, he argued.

He said he could not rule out that declining screening rates might explain the falling incidence rates. Still, he preferred the theory that E+P was a cancer “promoter.”

Reporters went gaga.

The New York Times treated Dr. Ravdin’s speculation as a solid theory despite not having access to the unpublished study or its data. An editorial in the Gray Lady gushed about the “astonishingly good news,” giving it instant credibility and causing worldwide panic.

Stock in Wyeth Pharmaceuticals (which competes with GlaxoSmithKline) tanked as menopausal women flushed their prescriptions down the toilet and reached for the ice packs.

During the media storm, I interviewed Dr. Ravdin’s co-author, Dr. Edwards. She was livid that he had gone public with a claim she said was not supported by data.

In her mind, the decline in mammography use accounted for the relative decline in incidence rates.

Dropping the usual caution of press-wary bureaucrats, Dr. Edwards darkly speculated that the media attention would result in a re-write of the study with a conclusion more in line with Dr. Ravdin’s anti-E+P spin. She sounded bitter and disappointed.

By the time the paper was published in The New England Journal of Medicine in 2007, Dr. Edwards’s fears had come to pass: the study asserted that the falling incidence rates were caused by a drop in the use of E+P.

The same edition of the journal also featured damning critiques from a dozen world-class epidemiologists who pointed out a raft of weaknesses in Dr. Ravdin’s causal theory. “We believe that most of the decline in the incidence of breast cancers is the result of screening,” several cancer specialists declared.

But the critiques were not headline generators.

Another paper that year, authored by Christina Clarke of the Cancer Prevention Institute of California and Anthony Robbins of the California Cancer Registry, likewise linked a drop in the use of hormone replacement therapy to a drop in breast cancer incidence.

The study, limited to middle-aged white women in California, tied falling breast cancer incidence rates between 2001 to 2003 to a statewide decline in the use of E+P after the bad news reports in 2002.

(It is worth noting that the temporary breast cancer incidence rate decrease actually began in 2001, a year before women abandoned Prempro.)

The authors speculated that Prempro somehow fueled breast cancers for white women. They claimed that when masses of menopausal white women suddenly went off the drug in 2002, some growing breast cancers instantly regressed.

They also denied that a drop in screening rates could be explanatory of the falling incidence rates, writing, “We found no evidence of a decline in reported use of screening mammography for the population studied.”

But according to the California Health Interview Survey, mammography use by middle-aged white women in the Bay Area fell by 2.6 percentage points between 2001 and 2003; statewide, the drop for that demographic was 1 percentage point. Screening rates dropped even more dramatically across the nation in 2003.

In Marin County, all white women received 5.6 percentage points fewer mammograms in 2003 than in 2001—a large drop in screening.

Even so, the decline in incidence reported by Dr. Clarke and Dr. Robbins was practically non-existent in absolute terms; the statewide incidence rate for middle-aged white women dropped by 29 cases per 100,000 from 2001 to 2004, resulting in a change of 29/1,000s of 1 percent.

But that did not stop the authors from framing the breast cancer incidence rate drop as an “unprecedented” 9 percent.

Dr. Clarke did not respond to requests for comment on this issue.

By 2009, Wyeth had faced 13,000 lawsuits alleging that Prempro had given users cancer, with juries awarding hundreds of millions in damages for the overselling of benefits from drugs concocted from or mimicking estrogen.

And lawyers hired experts like Dr. Clarke, who has disclosed that she receives consulting fees from plaintiffs’ attorneys litigating hormone therapy lawsuits.

Hormone replacement therapy lawsuits persist, even though the Women’s Health Initiative’s termination of the E+P trial is widely considered to have been a false alarm.

A steady stream of reputable studies have documented the benefits of hormone replacement therapies, including E+P and estrogen alone, and the non-associations of these drugs to breast cancer.

There are studies that say the opposite, too; it is best to check the funding sources of any study to identify the special interests involved, especially when researchers are on the payroll of competing pharmaceutical concerns.

Investigators with the Women’s Health Initiative have continued to massage data from the old E+P trial. In 2010, they discerned another tiny association between E+P and the risk of dying from breast cancer, claiming that using the drug may have caused an extra 1.3 breast cancer deaths per 10,000 women.

The investigators noted, however, that the miniscule blip of 1/100 of 1 percent could have been an artifact of incorrectly labeled death certificates, which are used to generate cancer mortality rates.

Closer to home, papers published by the Marin Women’s Study collaborative claimed that breast cancer incidence rates in 2003 fell in tandem with the discontinuation of the use of E+P after the shock of the health initiative news.

The county researchers argued that a higher-than-average use of E+P before 2002 could explain a good part of Marin’s historically high incidence rates; but since it is not possible to determine the rate of E+P cessation, the theory is no better than a trial balloon.

The falling breast cancer incidence rates in 2003 can much more easily be explained by simultaneously falling screening rates, as these same researchers have acknowledged in the scientific small print that reporters do not usually read. (The Pacific Sun, for example, published a story titled, “Breast Cancer Reduction Plunge in County Rates Linked to Plunge in Hormone Therapy, Say Medical Experts.”)

The powerful anti-hormone therapy meme that was unleashed by the Prempro trial stoppage and The New York Times’ promotion of Dr. Ravdin’s extravagant claim has uncannily persisted, despite an overwhelming absence of evidence—which, in the agonizing words of Donald Rumsfeld, “is not evidence of absence.”

On the other hand, there is plentiful evidence that ingesting sugar promotes cancer. But that is a story for another day.


Next week “Busted!” will explore the perils of relying on mammography as a public health measure.


A correction and a clarification: The opening quote by the Cancer Prevention Institute of California in last week’s installment of our series on breast cancer misstated the year: the statement was made in 2002, not 2012. Our apologies for the editing error. In the previous installment,  “The Marin Syndrome,” we discussed percentage change; we have since amended the article online to clarify that we meant percentage point change, which is not exactly the same thing.

Read part one here:

Read part two here:

Read part three here:

Read part four here:

Read part six here:

Read part seven here:

Read part eight here:

Read part nine here:

Read part ten here:

Read part eleven here: