National Cancer Institute Workshop Doesn't Work

By Joel Brind, Ph.D.

The opening address to the National Cancer Institute (NCI) workshop, conducted the last week of February and entitled "Early Reproductive Events and Breast Cancer," seemed auspicious. Dr. Andrew von Eschenbach, the NCI's director, gave a succinct history of the agency's "Abortion and Breast Cancer" fact sheet. (I refer to this as the ABC link.) However, his call for the workshop to "carefully, respectfully and vigorously discuss" the scientific data went unheeded.

The fact sheet first appeared on the NCI's web site (www.cancer.gov) in October 1994. It underwent six revisions, Dr. von Eschenbach noted, "but never with a comprehensive review of the data." He spoke of the most recent revision, which was posted only two days after he took office a year ago, and of the need "to rescind my approval of the fact sheet" and convene this workshop.

NRL News readers will recall that the information in that latest revision was, in fact, even less accurate than what it replaced. It went from labeling the evidence for the ABC link "inconsistent," to proclaiming that "it appears that there is no overall association" between abortion and breast cancer. The removal of the fact sheet from the web page, Dr. von Eschenbach said at the workshop, was prompted by "queries and challenges" that his office had received about its claims.

Those challenges had come in the form of a letter from the office of Rep. Chris Smith (R-NJ), signed by Smith and 27 of his fellow congressmen, and an analysis of the fact sheet published by the National Physicians Center for Family Resources. The congressmen's June 7, 2001, letter asked the Secretary of Health and Human Services (HHS) to "remove" and "reevaluate" the fact sheet, asserting that it was "scientifically inaccurate and misleading to the public."

In the months since the NCI removed the fact sheet (during which time it posted an interim message about "early reproductive factors," a vague term meant to include abortion without mentioning the word), there was considerable political pressure put upon the agency by pro-abortion congressmen and the media to restore the full-scale dismissal of the evidence of the ABC link.

Letters originating from the office of Rep. Henry Waxman (D-Ca.) and signed by 11 of his fellow pro-abortion colleagues referred to the ABC link as "popular myth" and asked for the restoration of the latest revised fact sheet. Predictably, sympathetic media rallied to Waxman's cause, accusing the Bush Administration of political meddling with "objective" science.

In early January, when the NCI announced the workshop, the New York Times threw down the proverbial gauntlet: "If the experts at the meeting agree that there is no link between abortion and breast cancer, the institute will have no excuse to suppress the information. It will have to issue a new fact sheet or admit it can no longer provide objective guidance on matters that inflame social conservatives."

In his introductory remarks Dr. von Eshenbach said that the data required the group's "scientific scrutiny." He charged the approximately 100 workshop participants to "carefully, respectfully and vigorously discuss" the scientific data.

Unfortunately, others clearly had different ideas. My own invitation to attend was secured only at the 11th hour, and even then, my participation was limited to being a member of the audience and a severely outnumbered panelist in the small-group "breakout session." Moreover, I was the lone dissenter.

In addition, discussion of the ABC link was severely limited, diluted by inclusion of research unrelated to induced abortion. Even Janet Daling, author of an important 1994 study which found a significant ABC link, was invited only to present data on breast tumor genetics, and not induced abortion.

In the only general session open to the public which did concern induced abortion, Dr. Leslie Bernstein of the University of Southern California Medical School did not critically review any previously published data. In a sign of things to come, she instead presented new data from one new study just published in February of this year, and her own group's study, not yet published.

In speaking of the previously published data, she took the 1997 Danish study by Mads Melbye et al.--which found no overall increased risk associated with induced abortion--at face value. She never addressed any of its flaws that had been discussed in the literature.

Dr. Bernstein essentially dismissed previous studies--the overwhelming majority of which did show evidence for the ABC link--as flawed. She attributed this to errors in defining the characteristics of the reference group to which the breast cancer patients are compared.

Specifically, she said "the majority of early studies didn't say who was in the reference group." I challenged this claim in the Q&A session which followed.

I pointed out that even the earliest study, published in Japan in 1957, very carefully matched breast cancer cases and healthy controls by parity (number of children). In other words, women with breast cancer (cases) were compared to healthy women (controls) who had the same number of children, so that the risk-lowering effect of having children did not confound the results. Dr. Bernstein replied that she'd meant problems with separating induced abortion from spontaneous abortion (miscarriage) in the early studies.

However, in 1996, when my research team and I assembled the worldwide data from the 28 extant studies in our comprehensive review and meta-analysis and found that induced abortion increases a woman's risk of breast cancer by 30%, we only included studies that specifically differentiated between induced and spontaneous abortion. (Spontaneous abortions do not increase the risk of breast cancer.)

I also asked Dr. Bernstein if the data from her new study would be made available at the workshop for our examination. She replied that she would not release any of the data until it was published!

There were two other presentations concerning the ABC link--short presentations under the heading "Late-Breaking Results" (i.e., data unavailable for scrutiny). The two presenters were Polly Newcomb, who had published a study on American women in 1996, and Mads Melbye himself.

Newcomb's earlier study had reported a 23% increased risk with abortion. But the new study on Wisconsin women (not yet published) showed zero increased risk! At least she did show the raw data briefly on the video screen--enough to show something very odd about how many women in Wisconsin had supposedly had an induced abortion.

During the Q&A I asked her to explain how less than 2.5% of women in her study had reported an abortion, while the national average is at least 10 times that rate. She said she did not know, speculating that the popularity of childbirth in Wisconsin led to a lower abortion rate than in other places.

Of course, the suggestion here is that her study may suffer from a massive degree of underreporting of abortions--a misclassification of women who've had an abortion as not having had any. But who knows? The study's details were not available for critical analysis.

The same troubling questions arose in the update of Dr. Melbye's 1997 study from Denmark. In following his cohort of Danish women from the year 1992 out to year 2000, he has found that there continues to be no overall association between induced abortion and increased breast cancer risk.

As noted previously in NRL News, although based on medical records, his study misclassified 60,000 Danish women who'd had abortions as not having had any. How?

By ignoring all the legal abortions that occurred in Denmark before 1973. I asked Dr. Melbye whether he had made an effort to find out about the thousands of abortions which had been misclassified. He said that such a task would be near impossible; these data are not on the Danish national computer registry.

So what he done in his follow-up? Readjusted his data to examine only the younger part of the population--for whom records were presumably complete--and again found no increased risk of breast cancer.

In other words, he superimposed a statistical adjustment to his flawed analysis instead of redoing the analysis and without including all the misclassified women whose abortions would have occurred before 1973. This operation can only yield the same extreme underestimate of the risk.

Those were the only opportunities during the entire two-day workshop for me to challenge any of the data contesting the ABC link. In the closed, small-group "breakout session," the group consistently voted that there is no association between induced abortion and breast cancer.

Of course I dissented. Most of the studies do, in fact, show an association. But the answer I got back was that there was "wide-spread agreement" within the group that no such link existed.

The same occurred in the final plenary session. All I could do was speak up from the audience and ask how the group could dismiss so many studies which did show an ABC link--especially since the more recent studies which concluded otherwise were subject to serious methodological questions and/or were just introduced at this meeting, with no chance for any critical analysis. Again, I got the non-answer about there being "widespread agreement" there was no link.

If the experts invited to the workshop had really wanted to settle the matter of the ABC link once and for all, to debunk the published evidence and relegate it to the scrap heap of epidemiological history, this was their chance to do it. I could have been allowed to present my analysis and they could have taken their best scientific shots for as long as they liked.

But they missed their chance. There was only one dissenting voice--namely mine--and I was given no podium to speak from, merely a seat in the gallery or around the table, and the chance to ask a question or two. The opportunity to raise serious doubts and challenges was never taken up.

There was no discussion of the four decades worth of published data that provided evidence for the ABC link, except, of course, for some of it that was updated by its own authors to erase their own earlier findings that did indicate an ABC link.

Yet there is some good news to report. While there was a refusal to acknowledge that having an abortion in and of itself increases a woman's chances of breast cancer, the flip-side, so to speak, was accepted. There was a set of important findings that unanimously acknowledged that a full-term pregnancy lowers the risk of breast cancer.

And although this protective effect is greatest for the first pregnancy, especially at a young age, it was found to apply to all subsequent full-term pregnancies as well. What does that mean?

For any woman already pregnant, choosing abortion will leave her with a greater long-term risk of breast cancer than she would have if she were to complete her pregnancy. In other words, as was even acknowledged in a new paper by John Thorp et al. in ObGyn Survey (a mainstream, peer-reviewed journal), induced abortion amounts to a risk factor for breast cancer because it abrogates the protective effect of full-term pregnancy.

Thorp et al. also acknowledge that women have a right to know that as a matter of proper informed consent.

The NCI can certainly make a good start by putting that information in a meaningful context for women considering abortion. And, oh yes, that "comprehensive review" of the data on the ABC link should be given a real opportunity to take place.