May cause cancer.

If there is a way to reduce the incidence of breast cancer, shouldn't American women be told about it?

Mr. Brind is a professor of biology and endocrinology at Baruch College of the City University of New York, where he has been teaching since 1986. His research on the connections between reproductive hormones and human disease has included breast cancer since 1982.

ON November 4 of last year, the Journal of the National Cancer Institute (NCI) published a study that the media have treated as an unscientific scare story, the way they should have treated the Alar and PCB cancer scares of years past. The study, by Dr. Janet Daling, et al., found a significant overall increase in breast cancer among Washington State women who had had one or more induced abortions (as opposed to spontaneous abortions, or miscarriages, which were not associated with increased risk).

The spin doctoring began immediately -- and not just in the popular media, but in the professional medical press. In fact, even as the Daling study rolled off the press, it was undercut by an accompanying editorial warning that "neither a coherent body of knowledge nor a convincing biologic mechanism has been established." Articles critical of the study continued to appear well into this summer.

To be sure, one ought to be extremely wary of raising public fears on the basis of any one study in isolation. This is especially important when one is considering such a high-incidence, potentially lethal disease as breast cancer (now estimated to strike about 12 per cent of American women), and such a high-incidence posited risk factor as induced abortion (over 1.5 million a year). If even the modest, 50 per cent overall risk increase found in the NCI study holds up, that would result in 40,000 to 50,000 additional cases of breast cancer a year, once the post - Roe v. Wade cohort starts to reach the age range at which breast cancer becomes more likely. Clearly, the fears engendered by such a study would affect a great many people.

On the other hand, if the study does prove valid, it is important to note that the posited risk factor is almost exclusively a matter of personal choice, and therefore avoidable in a way that environmental risk factors may not be.

In this case, however, the study is not isolated. Evidence of a possible connection between abortion and breast cancer has been published quietly since as far back as 1957.

The attack on the Daling study began, as I say, with a debunking editorial in the same issue of the NCI Journal, written by Boston epidemiologist Lynn Rosenberg. Her own work in the field, published by the American Journal of Epidemiology, was seriously flawed because the breast-cancer patients in her study were, on average, 12 years older than the cancer-free control patients. Since the risk of cancer increases with age, this was an egregious methodological error. Not surprisingly, her study evidenced a relative risk (RR) of breast cancer among women who had had one or more induced abortions of only 1.2 to 1.3 (i.e., a 20 to 30 per cent elevation in risk). This risk increase was reported as not statistically significant.

Perhaps that explains why her study was publishable in an American medical journal. The same American Journal of Epidemiology declined to publish an age-matched study of New York State women by Dr. Holly Howe, et al., of New York State's own Department of Health, a study which found a significant relative risk (RR) of 1.9 for breast cancer among women who had had any induced abortions. The Howe study was finally published in the English International Journal of Epidemiology in 1989.

While the American Journal of Epidemiology did not publish the Howe study, it did (in 1991) publish a comparison of studies done in Sweden which claimed to "explain the tendency toward increased risk of breast cancer which . . . appears to be associated with induced abortion" by a hypothesis called "response bias." According to this hypothesis, cancer-free women are more likely to deny abortions they had, while women with breast cancer are more likely to report their abortion history accurately. While the response-bias hypothesis is plausible and worthy of testing, the only evidence the study provided was that breast-cancer patients reported abortions of which the computer had no record. It is on this that the 1991 paper bases its assertion that patients tend to "overreport abortions" -- that is, to imagine abortions they never had.

Despite the absurdity of this evidence, the response-bias argument was mobilized to attack the Daling - NCI study. Harvard epidemiologist Karin Michels reported it to the New York Times as fact when the Daling study came out: "That [i.e., patient-recall-based data collection] is a flaw in the study design because women who had breast cancer are more likely to disclose an abortion than women who did not develop breast cancer." Lynn Rosenberg, in her editorial, suggested "the possibility of reporting bias" as a "limitation" of "major concern" in the Daling study, even though the Daling study itself soundly debunks the reporting-bias theory. So did the Howe study: while it revealed some misreporting of prior induced abortions either as being spontaneous abortions or as not having occurred at all, it also revealed that the misreporting "occurred similarly among the cases and the controls."

THE first study (in 1957) to show a significant association between induced abortion and breast cancer (relative risk=2.6) was performed and published in Japan, and subsequent reports with similar results were also published overseas. Another large, age- matched Japanese study (1982) showed risk to rise steadily with the number of induced abortions (RR=2.5 for one abortion, up to 4.9 for four or more). American studies showing significant risk increases among women on the West Coast (1981; RR=2.4 for abortion terminating a first pregnancy) and on the East Coast (the 1989 Howe study) were published in England. Similar findings from France (1984; RR=1.2 for one abortion, 1.6 for two or more), Denmark (1988; RR=3.9 for abortion terminating a first pregnancy), and the former Soviet Union (1978; RR=1.7 for any abortion) also popped up in European journals. In addition to the 10 epidemiological studies cited thus far, another 12 case - control studies have appeared in the peer-reviewed medical literature. Four (two in the United States and one each in France and Italy) showed no overall trend of increased risk (RR=0.9 to 1.1); three (one in America and two in Japan) showed risk elevations that did not achieve statistical significance (RR=1.2, 1.5, and 1.5, respectively); and four recent studies showed significant risk elevations, two in American women (RR=1.23 and 3.1), one in Greek women (RR=1.51), and one in Dutch women (RR=1.9). In fact, the only case - control study showing a negative association between induced abortion and breast cancer was a 1979 Yugoslavian study which was atypical in other ways as well. For example, it showed no evidence of the universally recognized protective effect of having children.

As late as 1992, the influential New England Journal of Medicine published an apparently comprehensive review of breast-cancer risk factors which inexplicably made no mention of the A-word, despite the fact that 13 out of the 14 case - control studies published by that time were consistent with increased risk.

In 1993, Harvard's Walter Willett, the epidemiologist among the four authors of the NEJM review, called this author "particularly sleazy in comparing a risk of breast cancer among women who elect to have an abortion with those who carry the pregnancy to full term." Willett's attack got curiouser and curiouser: "Of course the risk is higher among women having an abortion, not because abortions are a risk factor, but because a full-term pregnancy is protective." In fact, most studies correct for the protective effect of early full-term pregnancy. (Most studies also correct for such variables as number of children, use of oral contraceptives, and miscarriage.) They indicate that abortion further increases risk independently.

Earlier that year, the very same Dr. Willett had wasted no time in telling the world of a modest (though statistically significant) 65 to 85 per cent increased incidence of prostate cancer among men who had undergone vasectomy. The story was instantly picked up by the national media, even though the posited connection had theretofore been the subject of only a half-dozen epidemiological studies, with conflicting results. Subsequently, the subject was freely discussed in the Journal of the American Medical Association and other journals. Apparently, the possible risks of a minor surgical procedure for men is open for discussion in America. Not so for abortion.

Adding to the confusion is the fact that many studies over the years have failed to distinguish between induced and spontaneous abortion. This was true of 32 studies published between 1960 and 1994. Results of these studies have indeed been conflicting, some showing increased risk, some decreased risk, and some no effect. That is not surprising, given the difference in the effects of induced and spontaneous abortion.

The reasons for those differences are straightforward. The first trimester of a normal pregnancy is marked by a surge of hormones from the mother's ovaries, including progesterone, to maintain the pregnancy, and estrogen, which makes the breasts grow. Most known breast-cancer risk factors act via some form of overexposure to estrogen. Normally, the high estrogen levels of early pregnancy are counterbalanced by other hormones late in the pregnancy, which differentiate the breasts into milk-producing organs, thus rendering them permanently less susceptible to cancer. However, if the pregnancy is artificially terminated, the growth-stimulating effects of the estrogen surge help primitive and/or abnormal cells to grow into potential cancers. Contrariwise, as more than twenty years of research have shown, most first-trimester spontaneous abortions are characterized by subnormal secretion of ovarian hormones, including estrogen, whether because of inadequate stimulation by an abnormal fetus or because of an inadequate response by abnormal ovaries. Clearly, the failure to distinguish between spontaneous and induced abortion is a fatal weakness in any study.

Other confounding variables, such as socioeconomic class, race/ethnicity, and diet may also contribute to apparent increases in breast-cancer risk, and these are controlled for (sometimes well and sometimes not so well) by the selection of an appropriate control group. All in all, the best evidence for the real existence of a link between induced abortion and breast cancer is that it has been repeatedly observed in so many studies in different countries of widely varying ethnicity, diet and other lifestyle factors, and baseline breast-cancer incidence, and over a time span of almost four decades.

THE American medical media's wall of silence on the possible link between abortion and breast cancer was first breached in December 1993 in the African-American Journal of the National Medical Association. As Dr. Amelia Laing, et al., noted in the introduction to their age-matched Howard University study of over 1,000 black women, "Breast cancer is the leading cause of cancer mortality in black women," and "among women under the age of 50, the mortality rate in whites has declined, while it has increased in blacks." Thus (with black women also heavily overrepresented among abortion clients) there was a compelling interest in publishing their findings of significantly elevated risk among black women with any induced abortions (RR=2.7), risk which steadily rose with age until it reached 4.7 in the fifty-and-over group (which made up the majority of the study population).

The Howard University study was ignored by the mainstream popular press. Even the Philadelphia Inquirer, which heralded the Daling - NCI study, made no mention of the Howard study in a 3,000-word feature story on breast cancer in black women, written two months after the Daling - NCI story.

The media also have ignored three additional studies published this year that reconfirmed the significant findings of the Daling - NCI study. One of these, a study of women in Greece, constituted part of the doctoral thesis of Harvard epidemiologist Loren Lipworth. Among Lipworth's co-authors is the same Karin Michels who was interviewed about the Daling - NCI study by the New York Times, which reported: "Ms. Michels said she had reviewed in detail 40 published studies on abortion and breast cancer and had found no evidence of an increased risk." This story appeared one week after the Lipworth study she co-authored -- which confirmed Daling's finding of 50 per cent increased risk -- was submitted to the International Journal of Cancer.

Reporters are unquestionably in a difficult position when the people they should be able to turn to for accurate information refuse to provide it. Even Dr. Clark Heath, vice president of epidemiology and surveillance research at no less than the American Cancer Society, cited the unsubstantiated "reporting bias" argument when queried by a reporter about the Daling - NCI study, as reported by the Washington Times. He reacted as one hardly concerned, let alone alarmed, about a potential cancer risk that women might avoid, if given an informed choice: "Overall, it's a bit of a wash," he said.

One would have presumed that someone in Dr. Heath's position would be familiar with the considerable body of epidemiological data that has been gathered around the world for decades. One might also have expected him to point out a particularly disturbing fact in the Daling - NCI study: of the more than 1,600 women in the study, there were 12 who a) had had one or more abortions before age 18 and who b) also had a family history of breast cancer; all 12 were in the cancer-patient group, making the relative risk incalculably high for women with both of these risk factors. But that disturbing finding didn't make the news either.

Instead, denial from high places has even turned to outright disinformation. Dr. Devra Lee Davis, Senior Advisor to the Assistant Secretary for Health, juxtaposed, for the Washington Post (March 14, 1995), the long-standing legality of abortion in Japan with the fact that Japan has "the world's lowest breast-cancer rate." Dr. Davis thus suggested the absence of a connection between abortion and breast cancer. But all four Japanese epidemiological studies show a higher incidence of breast cancer among the small minority of Japanese women who have actually had any abortions.

Even more brazen was a piece written for the February 1995 Elle magazine by Assistant Surgeon General Susan Blumenthal, in which she said that "the [Daling - NCI] study did not consider the effect of birth-control pills," adding that "it is vital to rule out the effect of the Pill in any study of breast-cancer risks." It is also vital to get your facts straight. Not only did the Daling study "consider the effect of birth-control pills," but the possible effect of oral contraceptives on breast-cancer risk was a major focus of the study. My call to Dr. Blumenthal's office was returned by her assistant Teddy Fine, who doubted that a published correction was possible, since "the water is so far past the bridge at this point."

Yet five months further downstream, the pre-eminent professional journal, Science, skewered New York Times reporter Lawrence Altman for having given the Daling study any credence whatsoever back in 1994 (his story had run under the headline: "New Study Links Abortions and Increase in Breast Cancer Risk"). "Inevitably," said Charles Mann in the Science article, "public attention was directed to a risk that is unlikely to be real." Mann based this conclusion entirely on the assumption that Karin Michels was telling the truth. All Mann did was wrap her lie in more elegant imagery, as he called the Daling study "the one positive result in a sea of negative data." My polite corrective letter to the editor of Science drew only a form letter expressing "regret" over "space restrictions."

Most recently, on September 30, the literal eve of "Breast Cancer Awareness Month," the English medical journal The Lancet published an apparently comprehensive review article, "Breast Cancer: Cause and Prevention," by Drs. B. S. Hulka and A. T. Stark of the University of North Carolina. Claiming, "We focus on primary prevention," these authors somehow failed to mention the most obviously preventable risk factor for breast cancer: induced abortion.

One needn't look very far to find the motivation behind the increasingly desperate attempts to prevent public access to the considerable body of evidence of a connection between induced abortion and breast cancer: the reputation of abortion as safe for women is crucial to the "pro-choice" movement. The American Medical Association staunchly maintains that the risk of dying in childbirth is 12 times greater than the risk of dying from an abortion. Now, the risk of dying in childbirth is less than 5 in 100,000. If the overall increase in breast-cancer risk caused by induced abortion is even the modest 50 per cent suggested by numerous studies, that would raise lifetime risk from 12 per cent to 18 per cent -- an increased incidence of 6,000 per 100,000 women who have had any abortions. Even with a breast-cancer cure rate of 75 per cent, the increase in the death rate from induced abortion would calculate out to 1,500 per 100,000, making abortion 300 times more likely to result in death to a woman than childbirth. But mentioning that would be very un-PC.