By William J. Cromie

Gazette Staff

Available evidence is not strong enough to associate abortion with an increased risk of breast cancer, Harvard researchers say.

Their view contradicts the conclusions of a similar, recently published study claiming that abortions cause 4,700 excess cases of breast cancer each year among women in the United States.

As occasionally happens in science, two groups looked at similar data and came up with different conclusions.

Joel Brind of City University of New York analyzed results from 23 studies involving 61,000 women and judged abortion to be "a significant independent risk factor" for breast cancer.

Karin Michels and Walter Willett of the School of Public Health checked much the same data, and they decided that any such risk "is likely to be small or nonexistent."

The studies analyzed by Brind vary in quality. "They are inadequate to infer with confidence the relation between abortion and breast cancer," Michels says. "You obscure the issue if you combine them all and just look at an overall estimate, instead of trying to understand why the studies differ in their results."

For example, the studies used by Brind were mainly retrospective, that is, researchers asked women who already contracted breast cancer if they had undergone an abortion. "Some women might not admit this even to their mothers or husbands," Michels points out. "If women without breast cancer are more willing to volunteer information than those who are healthy, there will be a bias."

In other words, the results of a study can be skewed by the unwillingness of women to confide in researchers they do not know.

Then there's the tricky fact that a full-term pregnancy increases the short-term risk of breast cancer, but over the long term having children provides protection against the disease. "If a woman has an abortion, she misses out on the long-term protection, so the appearance of increased risk might come from this loss of protection," Michels notes. "This possibility was not considered in all the studies."

The added risk of breast cancer lasts a few years after pregnancy before it reverses. "Some of the studies, especially among younger women, did not follow the subject long enough to take this into account," Michels notes.

The short-term risk rises because levels of estrogen increase steeply, particularly during the first trimester of pregnancy. This hormone causes breast cells to multiply rapidly, a condition favoring cell mutation and cancer.

But anything that raises estrogen over many years gives postmenopausal women more protection, including pregnancies, early menarche, delayed menopause, and postmenopausal treatment with estrogen. These factors, however, "have only modest effects on breast cancer risk and indicate that a major influence from a few months of exposure to elevated [estrogen] would be unlikely," says Willett, Fredrick Stare Professor of Epidemiology and Nutrition.

Prospective vs. Retrospective

Michels and Willett also looked at prospective studies, wherein women who experience an abortion are followed to see if they develop breast cancer. "Such studies eliminate some of the potential biases of retrospective studies," Michels points out. The only such study of induced abortion found a decreased breast cancer risk.

Brind looked only at studies of induced abortion. Michels and Willett checked studies of both induced and spontaneous abortion. No link between breast cancer and spontaneous abortions was found in three large prospective studies.

"About half of the studies of induced abortion showed increased risk, but the rest concluded there was no risk, or a reduced risk," Michels explains.

Brind puts the increased risk at 30 percent. That's not much until you consider that an estimated 32.5 million American women have undergone abortions since 1973, when they became legal in the U.S. That figure increases by about 1.5 million every year. Brind calculates that, by the year 2030, abortions could produce 24,500 new cases of breast cancer.

"The evidence just doesn't justify that conclusion," Michels says flatly.

Further complicating the controversy is the fact that Brind has a record as an anti-abortion activist. He writes frequently on the subject, and admits to being sure abortion increases breast cancer risk before he did his analysis. In this situation, scientists often find what they're looking for.

"What is relevant is the science; the data should speak for themselves," Michels insists. "At this time, they tell us that they are not good enough to allow any valid conclusions about the association between induced abortions and breast cancer."

That doesn't prove Brind is wrong, however. "Maybe there is an increased risk," she admits. "But we will need more and better prospective studies to find that out."

Michels proposes that registries of abortions and breast cancer would be a good place to look. Such records are not kept in the U.S., but are available in Sweden and other Scandinavian countries.

Michels and Willett published their results in September in the journal Epidemiology. Brind reported his results earlier this month.