Biomarkers in Benign Breast Disease: Risk Factors for Breast Cancer

Benign breast diseases are both common and heterogeneous. Some types (usual and atypical hyperplasias) are weak but wellestablished risk factors (1) and probably precursors for breast cancer (2). Other types (e.g., sclerosing adenosis, papillomas, and fibroadenomas) also may be associated with a slightly elevated risk, although the evidence is less compelling (3ߝ5). These benign lesions are defined by their histologic features, which are insufficient to distinguish breast tissue destined to develop cancer. An article by Rohan et al. (6) in this issue of the Journal describes an interesting study that evaluates the abilities of certain biomarkers to better discriminate risk in patients with benign breast disease. Using a nested case-control study design based on the Canadian National Breast Screening Study (NBSS), Rohan et al. evaluated p53 and c-erbB-2 abnormalities by immunohistochemistry in benign breast tissue of patients who later developed breast cancer (cases; n471) compared with matched patients who did not (controls: n = 291 for c-erbB-2; n = 288 for p53). Alterations of the p53 tumor suppressor gene and c-erbB-2 oncogene are very common in breast cancer (7,8), which was the rationale for focusing on these markers as potential risk factors in this study. Most p53 abnormalities in breast cancer are missense point mutations resulting in nuclear accumulation of presumably nonfunctional protein (9). Most alterations of c-erbB-2 are gene amplifications that are highly associated with overexpression of the oncoprotein at the cell membrane (10,11). Immunochemistry was well suited to screening for p53 accumulation and c-erbB-2 overexpression in the archival tissue samples available in the study by Rohan et al. (6). Evaluating the risk of a relatively rare disease (e.g., breast cancer) in a relatively healthy population (e.g., patients with benign breast disease) can be difficult for a number of reasons, and the nested case-control design used in the study by Rohan et al. was an effective strategy to compensate for some of these difficulties (12). By sampling controls matched to breast cancer cases from the entire cohort of all NBSS enrollees with benign breast disease, the study had power nearly equivalent to that of the entire cohort (n = 4888) but evaluated far fewer subjects (n = 359). Nuclear accumulation of p53 in benign epithelial cells was associated with a statistically significant increased risk of developing breast cancer in the study by Rohan et al. (6) (adjusted odds ratio 4 2.55). The median interval between diagnosis ofbenign disease and subsequent cancer was only 2.1 years, suggesting that p53 accumulation may be a marker for rapidly developing tumors or, less likely, for clinically occult breast cancers. Apparently even weak focal immunostaining in any type of epithelium, including histologically normal cells, was predictive for cancer. This weak immunostaining is surprising in the sense that mutationally inactivated p53 is associated usually with strong diffuse protein accumulation (13), and most previous studies (14) have found no convincing evidence of p53 mutations in normal or benign breast epithelium. However, low levels of p53 may accumulate in any type of cell in the absence of mutations, such as during the G1/S interphase of the cell cycle in response to DNA damage or oxidative stress, which presumably allows time for the repair of damaged DNA or shunting of severely damaged cells toward apoptosis (15). Thus, low levels of p53 protein may be a marker of cells exposed to a carcinogenic microenvironment, leading to increasing genetic instability, tumor initiation, and disease progression. Perhaps larger studies will disclose whether p53 accumulation adds substantial additional risk to patients with specific types of benign lesions such as usual and atypical hyperplasias. Overexpression of c-erbB-2 in benign breast tissue was not predictive for developing breast cancer in the study by Rohan et al., although there was a statistically nonsignificant trend toward lower risk in positive cases (adjusted odds ratio 4 0.65). Positivity for c-erbB-2 was defined as any immunostaining involving the surface membranes or cytoplasm of epithelial cells, and cytoplasmic staining alone accounted for the majority of positive cases. The prognostic implications of c-erbB-2 overexpression in breast cancer are overwhelmingly in the direction of unfavorable outcome related to membrane staining (8), and the few studies addressing cytoplasmic staining have discordant results (16ߝ18), which cautions against attaching too much significance to the favorable trend observed in this study. In addition, most previous investigations (14) have found no evidence of substantial c-erbB-2 overexpression in normal or benign breast epithe-lium, and gene amplification and messenger RNA expression of c-erbB-2 appears to be associated with membrane rather than cytoplasmic immunostaining (19ߝ21). Certain antibodies, including the one used in the study by Rohan et al. (6) (NCLCB11), are prone to cytoplasmic immunostaining, which can mimic a weak membrane signal in closely apposed cells; there also is evidence that cytoplasmic staining may represent a crossreaction with a mitochondrial protein unrelated to c-erbB-2 (22). These potential methodologic problems may have contributed to the puzzling and essentially negative findings regarding cerbB- 2 in the Rohan et al. study, and the issue deserves further consideration. Because these types of studies are so difficult, there are few of them to compare. In addition to the study by Rohan et al. (6), perhaps the only other substantive similar investigation also found that alterations of p53, among several other factors including c-erbB-2, were associated with additional hazard for developing malignant disease (23). Because the studies are few and small, their findings need to be validated in larger numbers of patients with longer follow-up time. Overall, by themselves, the results of the study by Rohan et al. (6) are not strong enough to alter how we treat patients with benign breast disease. The absolute risk for developing breast cancer associated with p53 accumulation can be estimated by back-calculating the adjusted odds ratios observed in the study to the hypothetical cohort that generated the case subjects and matched control subjects. These mathematically derived estimates are 3.3% for patients with p53-positive benign breast disease and 1.3% for patients with negative p53 status. The vast majority (96.7%) of women with p53-positive lesions did not develop cancer within the time frame of the study, and the vast majority (86%) of women who developed cancer did not have p53-positive benign disease. Nonetheless, the study is noteworthy in identifying p53 as a potential risk factor and, more importantly, as a paradigm for future studies of additional biomarkers that together may eventually identify patients with high-risk benign breast disease who might benefit from closer surveillance, surgical intervention, or chemoprevention.