In a large, formal Philadelphia courtroom six years ago, endocrinologist Joel Brind swore on a Bible, took the witness stand—and forever left behind his life of scientific obscurity. Brind was the star witness for Christ's Bride Ministries, a religious group that had used billboard space throughout the Northeast to make the claim that abortion increased a woman's chance of developing breast cancer. The Southeastern Pennsylvania Transportation Authority, which runs the city's subway, bus, and commuter rail systems, rejected the advertisements as scientifically unsound, and the whole matter landed in court.
Brind, who teaches human biology and endocrinology at Baruch College of the City University of New York, made an impressive expert witness. A lanky 52-year-old with a slender, equine face, he has spent most of his career investigating the connections between reproductive hormones and human disease. He has a talent for explaining science in a patient way that a layperson can easily understand. On the witness stand that warm June morning, he unveiled a theory that he was about to publish in a British public health journal called Epidemiology and Community Health.
"Within a few days after pregnancy, the corpus luteum, which is in a woman's ovary, begins to secrete large quantities of a number of hormones," Brind told the crowded courtroom. One of those chemicals, estrogen, makes the breasts grow in preparation for nursing. In the early months of a first pregnancy, "the breasts may be adult size, but the tissue is rather primitive. In other words, it's not specialized for producing milk. It's mostly able just to grow, to proliferate." Later in the pregnancy, he said, the growth switch clicks off, and those cells differentiate into mature, milk-producing cells.
"Now, primitive cells, because they're programmed to grow, are more likely to be sensitive to carcinogenic stimuli," Brind said. If a woman has an abortion, she's left with a large number of these immature cells lining her breast ducts, and she is therefore more vulnerable to cancer down the road—30 percent more vulnerable, Brind says, than a woman who has never had an abortion.
This, the endocrinologist believes, is a terrible truth that the medical establishment—including the National Cancer Institute, the American Cancer Society and the World Health Organization—has tried to keep from the public. In the courtroom, James Owens, one of the attorneys for Christ's Bride, tried to imply a conspiracy of silence. "Do you have an opinion, Dr. Brind, as to why the link between induced abortion and breast cancer has not been promulgated in the public-health sector in the form of a warning to women who would choose abortion?" the lawyer asked. "Well, to use a common expression," Brind responded, "it would seem to be politically incorrect."
Ever since he testified in the Philadelphia case—which Christ's Bride Ministries won on free-speech rather than scientific grounds—Brind has spent about 90 percent of his time outside the classroom investigating and publicizing the link between abortion and breast cancer. He has testified in courthouses and statehouses in Arizona, Florida, Massachusetts, Ohio, North Dakota, New Hampshire, and Alaska; he has lobbied Congress and the Food and Drug Administration and written letters to scientific journals. His efforts are paying off: At least 18 states have considered laws requiring clinics to disclose the link, and Mississippi and Montana have passed such measures. The state of Florida has cited Brind's research in its efforts to require that parents be notified before minor daughters have abortions. Pro-life activists have used Brind's statements to defend billboard and television ad campaigns from Maryland to California. Recently, an antiabortion member of Congress, citing Brind's work, convinced the National Cancer Institute to dilute its long-standing assertion that women face little cancer risk by terminating pregnancies.
There's only one thing wrong with this picture: The vast majority of epidemiologists say Brind's conclusions are dead wrong. They say he conducted an unsound analysis based on incomplete data and drew conclusions that meshed with his own pro-life views. They say that epidemiology, the study of diseases in populations, is an inexact science that requires practitioners to look critically at their own work, searching for factors that might corrupt the results and drawing conclusions only when they see strong and consistent evidence. "Circumspection, unfortunately, is what you have to do to practice epidemiology," says Polly Newcomb, a researcher at the Fred Hutchinson Cancer Research Center in Seattle. "That's something Brind is incapable of doing. He has such a strong prior belief in the association [between abortion and cancer] that he just can't evaluate the data critically."
For Newcomb and many others, Brind's crusade highlights challenges that face those who are trying to understand the origins of diseases like cancer, where there isn't necessarily a straight line from A to Z. The crusade also serves as a warning about what happens when politics drives science. For laypeople who are trying to make sense of scientific controversies, it's a reminder of why it's important to study the research itself rather than simply trust the pronouncements of experts.
Joel Brind was a 10-year-old growing up in laurelton, New York, when he realized he wanted to become a cancer researcher. In the early 1960s, Life magazine published an article about the inner workings of cells, a world suddenly opened up by electron microscopy. "They said how scientists will figure out what the cells do, and they'll ultimately be able to cure cancer and stuff," Brind recalls. "I just decided, 'Wow, that's for me.'" Even before receiving his Ph.D. from New York University, he found himself studying sex hormones—in particular, androgens and their relationship to acne. Eventually, he began looking at estrogen metabolite levels in the blood, correlating them with the development of both malignant and benign breast diseases.
One day in late 1992, Brind was sitting at his dining room table in Matamoras, Pennsylvania, reading Science News, when he stumbled upon an article arguing that adolescent girls who get pregnant "seem to have a built-in defense against breast cancer later in life." The article did not mention whether the girls had to give birth to reap the benefit. Brind found himself reading the article three times through, thinking, "What happens if the teen terminates her pregnancy?" He started combing the library at the Mount Sinai School of Medicine in New York. Although 40 years of studies on the subject had yielded inconsistent results, he became convinced that abortion was a significant risk factor for breast cancer. It made sense physiologically, he thought: More than a decade earlier, biologists Jose and Irma Russo, then at the Michigan Cancer Foundation, had induced breast cancer in rats by aborting their pregnancies—leaving them with immature mammary gland cells—and exposing them to the toxic chemical 7,12-dimethylbenz(a)anthracene. The chemical was known to be particularly effective in producing mammary carcinomas in rats that had never given birth.
When he finished reading the articles, Brind could barely contain himself. He printed up some flyers with the headline, "Women Have a Right to Know," piled his wife and 9-year-old daughter into his Dodge station wagon, and drove to Washington, D.C., where he tried to meet with legislators on Capitol Hill. Sitting with his family in a congressional cafeteria, Brind spotted Representative Richard Gephardt and sent his daughter over to the congressman to hand him the leaflet. When Gephardt tried to walk away, Brind recalls, the little girl implored, "No, you really have to read this. My daddy says it's very important. An abortion can cause breast cancer."
Brind's peers told him he wouldn't be taken seriously unless he published his own analysis in a peer-reviewed journal. So he contacted two colleagues, Walter Severs and Joan Summy-Long, both endocrinologists at the Pennsylvania State College of Medicine and opponents of abortion. Since none of them were epidemiologists, the trio lacked the training to do a serious analysis of existing research. They approached Vernon Chinchilli, a biostatistician at the college. "When I heard what the topic was, red flags went off in my head," says Chinchilli, who describes himself as pro-choice. However, he agreed to hear out the threesome, and after several meetings, he decided to sign on to their team.
Brind spent the next year unearthing studies that correlated induced abortion (as compared with spontaneous abortion, or miscarriage) with breast cancer. His thoroughness in sifting through the literature worldwide, in several languages, allayed many of Chinchilli's concerns. "I would have never gone through the torture he did," the statistician says. To avoid accusations of bias, the researchers decided to include every relevant study they could locate—28—although some contained outdated methodology. "We realized that because it was such a contentious area of research, the safest thing to do would be to be the most inclusive—to include everything," Brind says. Then Chinchilli plugged their numbers into a computer.
Brind was at home the day his fax machine spit out the results of Chinchilli's analysis: Women who had induced abortions appeared to suffer from breast cancer at a rate 30 percent higher than those who didn't. In biostatistical terms, this is called a "relative risk" or "odds ratio" of 1.3. (A risk of 1.0 means there's no danger.) Because there are so many variables that can corrupt data, epidemiologists regard any risk under 2.0 with skepticism unless it appears consistently from one study to the next. (For example, scientists at the National Cancer Institute once argued that mouthwash users had a 50 percent higher chance of contracting oral cancer. Critics later showed that once other variables such as alcohol and tobacco use were properly controlled for, the risk all but disappeared.)
In later meetings, Chinchilli tried to control the enthusiasm of the three pro-life scientists. "Being a statistician, I have some doubts," he says. "I don't think the issue has been resolved." But Brind remained resolute in his conclusion that the abortion-breast cancer link had been proved. "When we were talking about the conclusions, he wanted to make the strongest statements," Chinchilli recalls. "I tried to temper them a little bit, but Dr. Brind is very adamant about his opinion."
If Chinchilli was cautious in pronouncing a link between abortion and breast cancer, other scientists were—and still are—dismissive. "In epidemiology, if there's a true relationship, you'd expect to find the majority of studies would show some consistency," says Phyllis Wingo, chief of the cancer surveillance branch for the Centers for Disease Control and Prevention. When Wingo surveyed the literature for the journal Cancer Causes and Control, she found results that were all over the map. "There were some studies that showed a small increase in risk, some that showed a small decrease, and some that showed neither. That lack of consistency was not persuasive." Besides, she says, a relative risk of 1.3—compared with the relative risk of 20 associated with smoking and lung cancer—is usually considered too weak to draw definite conclusions.
One conclusion researchers have drawn is that the poor design of many abortion-cancer studies produced results that exaggerated risks faced by women who undergo the procedure. Until recently, most of the published papers on the subject were based on what epidemiologists call case-control studies. Researchers would find a group of cases—women diagnosed with breast cancer—and ask them whether they had had abortions earlier in their lives. They would also survey a group of controls—healthy women who were contacted by random telephone dialing or some other method. The scientists would adjust for variables such as age, reproductive history, and family health history, then perform a calculation to determine the relative risk of having an abortion.
One variable, though, that epidemiologists can't eliminate is whether a woman admits to a researcher that she has had an abortion—because such an admission is still a source of shame for many people. This is the point at which results are most likely to go awry. By contrast, patients are always looking for clues to their illnesses, and so "women who have breast cancer will search their souls, and be very likely to search deeply in their memories, and disclose things that might be embarrassing," says David Grimes, clinical professor of obstetrics, gynecology, and epidemiology at the University of North Carolina School of Medicine in Chapel Hill. "But a woman who does not have the disease and is picked at random from the community is very unlikely to disclose to an anonymous researcher knocking on the door that she had an abortion in 1992." Epidemiologists call this phenomenon response bias, and they've found evidence that it can contaminate studies on abortion and breast cancer.
Brind dismisses response bias as an unproven hypothesis, but others have found ample evidence for it. In Sweden, epidemiologist Britt-Marie Lindefors-Harris of the Karolinska Institute took advantage of her country's nationwide registry of legal abortions. In a project documented in the American Journal of Epidemiology, Lindefors-Harris conducted a case-control study of abortion and breast cancer, but with a twist: She checked government records to see if the participants were telling the truth about their reproductive histories. Many of them, it turns out, were not. Out of 829 women, 29 appeared to misrepresent their abortion history, with the vast majority of underreporting coming from healthy women in the control group. Based on those numbers, Lindefors-Harris calculated that "an observed increase in risk of up to 50 percent may be caused by response bias."
Five years later, in 1996, Matti Rookus and Flora van Leeuwen of the Netherlands Cancer Institute came up with even more dramatic evidence of bias. The epidemiologists surveyed women in two regions of their country. In the liberal west, Rookus and van Leeuwen found a statistically insignificant relative risk of 1.3—but in the predominantly Roman Catholic southeast, relative risk shot up to an astounding 14.6. The only plausible explanation: Because of the conservative religious values in the southeast, healthy women there lied about their abortions. "Reporting bias is a real problem," the Dutch team concluded.
In January 1997, three months after Brind's analysis appeared in Epidemiology and Community Health, a Danish epidemiologist published a paper in The New England Journal of Medicine that, according to many of his colleagues, made much previous research on the subject moot. Mads Melbye of the Statens Serum Institute in Copenhagen, Denmark, looked at the records of 1.5 million women born in his country between 1935 and 1978. Like neighboring Sweden, Denmark records all abortions; it also has a national cancer registry. Melbye linked these two databases—and found a relative risk of 1.00. In other words, women who underwent abortions developed breast cancer at exactly the same rate as women who didn't. (Melbye did find a risk of almost 1.9 for women who had abortions in the 18th week or later, but such procedures are rare and done only in emergencies.) With so large a population and no opportunity for response bias, Melbye's study convinced many scientists that earlier case-control studies had been tainted. "The story of abortion and breast cancer is essentially closed by this last study," says Carlo La Vecchia, an epidemiologist at the Negri Mario Institute for Pharmacological Research in Milan, Italy.
"I think the Melbye study pretty much clinches it," says Lynn Rosenberg, associate director of the Slone Epidemiology Center at Boston University. "The numbers were huge. The larger the study, the more statistically powerful it is, and the more stable the results are."
Public health organizations across North America and Western Europe have downplayed the danger of terminating a pregnancy, but last summer the National Cancer Institute suddenly shifted gears. The institute, a research agency of the federal government, had published a fact sheet in March that said, "The current body of scientific evidence suggests that women who have had either induced or spontaneous abortions have the same risk as other women for developing breast cancer." This statement was withdrawn from the agency's Web site in July 2002 after Representative Christopher Smith, cochairman of the House Pro-Life Caucus, wrote to Secretary of Health and Human Services Tommy Thompson, calling the fact sheet "scientifically inaccurate and misleading to the public." In particular, Smith criticized the institute for failing to cite Brind's work, which he called "the only comprehensive review and meta-analysis" of the abortion-cancer link. In November the institute said the information had been taken down "to review it for accuracy," adding that Smith's letter "was a factor" in the decision. The institute has issued an interim statement that calls the data "inconsistent." Other credible agencies have been less reticent about dismissing the link outright. "Results from epidemiological studies are reassuring in that they show no consistent effect of first-trimester induced abortion upon a woman's risk of breast cancer later in life," says a June 2000 fact sheet posted on the World Health Organization's website.
Such pronouncements have only stoked Brind's crusade. He calls Melbye's study "horrible," saying the Danish epidemiologist "breaks almost every rule in the book" by manipulating data. (Brind has criticized how Melbye adjusted for the age of participants in his study. Other epidemiologists say Melbye was simply using appropriate statistical methods. Julie Palmer, an epidemiologist at the Slone Epidemiology Center, has testified that in Melbye's study "women of like ages were compared with women of like ages. It's a given that in epidemiologic studies, especially involving cancer, you must control for age.") Brind reserves his harshest words for groups such as the National Cancer Institute, which until its recent equivocation spent almost a decade reassuring the public that abortion appeared not to trigger breast cancer. Brind has charged scientists and staff there with engaging in "a miscarriage of scientific justice" and a "cover-up" of the truth. He has traveled the world spreading that message. In a speech in Australia, he accused the National Cancer Institute of "protecting the abortion industry by invoking flawed analyses from Sweden, the Netherlands, and Denmark."
To epidemiologist Mads Melbye, Brind's campaign is a failure of scientific objectivity. "I see this as an input from an antiabortion angle, which is very unusual for a Scandinavian," Melbye says. "Here, we want to be very objective and discuss science, and not religion."
'Humans Are Not Rats'
When Joel Brind tries to link abortion and breast cancer, he starts with Jose and Irma Russo's research. In 1978 the husband-and-wife team took 49 female rats and divided them into groups: rats that had carried a pregnancy to term and delivered offspring, rats that had been given abortions, and rats that had not mated. Then they exposed the animals to a chemical that is known to give female rats cancer: 7,12-dimethylbenz(a)anthracene. One of the 18 rats that had given birth developed a malignancy, seven of the nine rats that had abortions developed malignancies, and 15 of the 22 virgins developed malignancies.
Writing in The American Journal of Pathology in August 1980, the Russos concluded that a full pregnancy appeared to give female rats protection against a known carcinogen. They pointed out that female rats are born with terminal end buds: bulbous groups of cells at the tips of their mammary ducts that are prone to cancer. As pregnancy progresses, those cells differentiate into more mature structures called alveolar buds and lobules to prepare for nursing. The Russos drew parallels between rats and humans: Like rats, pregnant women undergo hormonal changes that stimulate the growth and subsequent specialization of breast cells into lactating cells. "Abortion would interrupt this process," they added, "leaving in the gland undifferentiated structures like those observed in the rat mammary gland, which could render the gland again susceptible to carcinogenesis."
Gil Mor, director of the reproductive immunology unit at the Yale University School of Medicine, says flat out: "Humans are not rats." Humans and rats are fundamentally different organisms, he says, pointing out that rats don't even have breasts and, therefore, "there is no breast cancer in rats. We [use] the rat to understand basic biological process. Period. Basic biological processes." In short, Mor says the Russos are on solid ground studying the basic processes of mammary-gland differentiation in rats. But when they or someone like Brind tries to extrapolate those processes to humans, the terrain gets wobbly.
Epidemiologists say that Brind has failed to establish a convincing link between abortion and breast cancer. "The first step in an epidemiological study is that you establish a consistent, believable association, and then you see if that association can be supported by the biological literature," says Eugenia Calle, director of analytic epidemiology for the American Cancer Society. "There is no consistent, believable association between abortion and breast cancer. So to spend our time trying to explain an association that we don't see is a little bit odd."