April 2009
Feature Story

Karen Lusky

Might pathologists one day routinely examine a benign breast biopsy to quantify the patient’s future risk of breast cancer? That’s the idea behind efforts at Mayo Clinic to develop a tissue-based risk prediction model, an initiative that could also shed light on the pathogenesis of breast cancer and ways to prevent it.

Mayo researchers have so far identified what appears to be an independent risk feature: breast lobule involution, a normal, age-related process in which the milk-making glands shrink. According to an analysis of almost 9,000 benign breast biopsies at Mayo, women with complete lobular involution at the time of their biopsy had a definite statistical edge in dodging breast cancer during a 17-year median followup period. The women’s biopsies were categorized as having no involution, partial (one percent to 74 percent), or complete (75 percent or more) involution.

The ongoing research, which lead investigator Lynn Hartmann, MD, a Mayo oncologist, reported at the December 2008 San Antonio Breast Cancer Symposium, is the first to find the association between involution and breast cancer. The phenomenon of involution as a risk factor is “independent of histology and family history,” Dr. Hartmann said.

The study findings debunk the notion that breast cancer risk steadily rises as women age. Instead, says Dr. Hartmann, the heightened risk appears to be concentrated in women over age 55 who have not gone through the involution process. Their relative risk is three times that of women in the same age group who have completed involution, she told CAP TODAY (Milanese TR, et al. J Natl Cancer Inst. 2006;98[22]:1600–1607).

Based on raw data that Mayo has compiled, she adds, it appears that women in the study with no involution were more likely to develop higher-grade cancers than women who had complete involution. “But this could represent an effect of the age of the women [younger] with no involution,” she says. “These additional analyses have not been completed.”

One cause for excitement about the Mayo findings, says the study’s pathologist Daniel W. Visscher, MD, professor of pathology at the University of Michigan, is that patients with complete involution and nonproliferative histology on their benign biopsies actually were found to have a lower risk for breast cancer than the female population overall. All other studies, he points out, have found that women with benign findings had either the same or a higher risk for breast cancer than the general population. Thus, involution appears to be a protective factor.

In fact, the hypothesis for the Mayo involution study, Dr. Visscher says, was driven by epidemiological data showing that breast cancer incidence starts to fall off at the time that involution should, in theory, be complete. The involution process starts in the 20s and ends sometime in the seventh decade for most women, though the timetable varies from woman to woman.

In the Mayo study, Dr. Hartmann says, “half of women in their 60s had completed involution. Only two percent had no involution—and those are the higher risk people.”

A study by Harvard researchers published early this year in Cancer also found a link between lobule type and breast cancer risk. The nested case-control study (200 cases, 915 controls) looked at benign breast biopsies and subsequent breast cancer in women enrolled in the Nurses’ Health Study. Pathologists reviewed the biopsy slides and counted the number of acini in the lobules to classify the lobules into types one, two, or three indicating the stages of morphologic development. The average time since biopsy was 9.7 years (Baer HJ, et al. Cancer. 2009;115[7]:1404–1411).

“During the involution process, type two and type three lobules regress back to type one lobules,” the study report says. Senior study author Rulla Tamimi, ScD, assistant professor of medicine at Harvard Medical School, Brigham and Women’s Hospital, says the study found that those with predominant type one lobules and no type three lobules had an approximately 30 percent reduced risk of subsequent breast cancer compared to women with other lobule types.

It’s believed, Dr. Tamimi says, “that a woman starts with the type one lobule, which is predominant in girls and women before they have had their first child. As women develop and go through pregnancy, they are more likely to have type three lobules, and as they age, their lobules regress and revert back to type one lobules.”

Mayo has also conducted a small case-control study in which Mayo pathologists quantified involution by counting acini in the lobules. In that study, involution “quite significantly” outperformed the Gail model in predicting risk of breast cancer, Dr. Hartmann says. “We have a manuscript under review at this time on that topic,” she adds. Mayo’s Kevin P. McKian, MD, presented the findings in a poster session at the December breast symposium.

Though it makes intuitive sense that breasts with atrophied lobules would be less likely to develop breast cancer, researchers haven’t yet pinpointed the reason. One theory, Dr. Visscher says, is that women who don’t go through the process end up having their breast cells exposed to carcinogenic influences for a longer period. An analogy might be the way in which endometrial hyperplasia develops in response to persistent or high estrogen levels, he notes.

“Another theory could be that failure to undergo involution is an early sign of biological tumor progression,” Dr. Visscher adds. “Maybe the persistence of the cells is a sign of a proliferative phenotype,” reflecting expression of traits favoring neoplasm. Breast cancer risk factors, such as atypia, are often constitutional, conferring “bilateral risk, as though all cells are predisposed,” he says.

Mayo researchers have found that some known breast cancer risk factors did not correlate to involution. For example, the researchers collected data on whether the women had “never or ever” used hormonal replacement therapy. And “the ‘ever users,’” Dr. Hartmann says, “were slightly less likely to have completed the involution process.” The difference between the never and ever users wasn’t statistically significant, however, though there was a trend.

As for the impact of a woman’s childbearing history, the more pregnancies a woman had, the more likely she was to retain the lobular structure, Dr. Hartmann says. Yet overall, pregnancy provides a slight protection against developing breast cancer. And however it mediates that effect, she adds, it’s not through involution.

The Mayo study, which looked at the women’s involution status only at the time of their benign breast biopsies, did not answer the question about whether involution might reverse itself in some women. “We are assuming,” Dr. Visscher says, “that involution progresses since we know the process is associated with age.” But “we don’t know what happens to involutional lobules if a woman gets pregnant later in life or takes estrogen or progesterone.” If involution does reverse itself or go back and forth, “one could speculate that might be a factor associated with developing breast cancer.”

As far as using involution for risk assessment, Dr. Visscher thinks it may eventually be possible to add a patient’s involution status to routine pathological observations for benign breast biopsies. The value of involution as a predictive marker, however, might be in combination with radiographic and clinical assessments, he says.

He points, for example, to “considerable evidence in the radiographic literature suggesting that breast density is related to breast cancer risk.” And, Dr. Hartmann says, “there is some correlation between involution and breast density on mammography because a woman with fatty replacement of her breast tissue will show low density on a mammogram. And these women are more likely to have complete involution.”

But breast density on mammography can represent just about any finding except fatty tissue, she adds, because “the little lobules [as they undergo involution] are initially replaced with stroma, which shows up dense on a mammogram.”

Dr. Hartmann and colleagues plan to take a look at whether combining a finding of density on mammography and histologically determined involution status provides better risk prediction than either feature alone. “We have mammography density data for almost 3,000 of the women in our cohort” of women with benign breast disease, Dr. Hartmann says, “so we are looking at that very question.”

Combining a woman’s involution status with atypia seen in a benign breast biopsy could also provide a more synergistic risk picture. That’s because “involution is associated with reduced risk in atypia,” Dr. Hartmann says. “The other thing that affects risk associated with atypia is the number of foci of atypia, which is also something pathologists can tell us but it isn’t done routinely at this time.”

Research done at Mayo and published in 2007 found that if a woman’s benign breast biopsy has only one focus of atypia, her relative risk for developing breast cancer is 2.3, Dr. Hartmann says (Degnim AC, et al. J Clin Oncol. 2007;25[19]:2671–2677). With two foci, it jumps to 5.3, and with three or more foci of atypia, the relative risk is eight times the risk of age-matched women. “That information,” Dr. Hartmann says, “perhaps by itself or in combination with involution, is going to be far better able to risk stratify these women.” (To put the risk into perspective, Dr. Hartmann notes that women with BRCA mutation have a relative risk 15 times higher. Hormonal replacement therapy confers a relative risk of 1.26.)

Mayo researchers have not yet come up with a risk calculation combining the number of foci of atypia with involution.

Dr. Hartmann and her research colleagues are also identifying other risk-related biomarkers, one of which is COX-2 expression in the breast tissue of women with atypia. A Mayo study published last year found overexpression of COX-2 was statistically significantly associated with the type of atypia, the number of foci of atypia, and age at the time of biopsy (older than 45 years). The study authors conclude that COX-2 could be a target for chemoprevention in the form of nonsteroidal anti-inflammatory drugs (Visscher DW, et al. J Natl Cancer Inst. 2008;100[6]:421–427).

Dr. Hartmann says Mayo isn’t ready to release other risk factors and biomarkers into the public domain.

“The whole point is to stratify risk into meaningful groups of high, moderate, and low,” she says, and then look for ways to determine who would benefit from breast MRI, for instance. And that might, for example, be women with atypia with no involution or those with atypia with multiple foci, she notes.

Says Dr. Visscher, “If we can identify women at very high risk” comparable to that conferred by BRCA mutations, those women “may want to get additional screening and perhaps pharmacological or surgical risk reduction.”