To better understand early steps in human breast carcinogenesis, we examined allele imbalance or loss of heterozygosity (LOH), in co-existing normal-appearing breast epithelium and cancers. We microdissected a total of 173 histologically normal ducts or terminal ductolobular units (TDLUs) and malignant epithelial samples from 18 breast cancer cases, and examined their DNA for LOH at 21 microsatellite markers on 10 chromosome arms. Fourteen of 109 (13%) normal ducts/TDLUs, from 8 of 18 (44%) cases, contained LOH. The location of these 14 ducts/TDLUs appeared unrelated to distance from the cancer. LOH in normal-appearing epithelium involved only single markers, whereas LOH in cancers commonly encompassed all informative markers on a chromosome arm. In only 1 of 14 (7%) ducts/TDLUs with LOH, was the same LOH seen in the co-existing cancer. Global differences in LOH per arm in normal-appearing tissue were not demonstrated, but less LOH was seen at 11q and 17p than at 1q (P ‫؍‬ 0.002), 16q (P ‫؍‬ 0.01), and possibly 17q (P ‫؍‬ 0.06). These results indicate that in a large fraction of women with breast cancer, histologically normal breast epithelium harbors occult aberrant clones. Individual clones rarely are precursors of co-existing cancers. However, they might constitute a reservoir from which proliferative lesions or second cancers develop once additional genetic abnormalities occur, they could contribute to intratumoral genetic heterogeneity, and they are consistent with a role for genetic instability early in tumorigenesis. Breast cancers, even carcinoma in situ (CIS), the earliest recognized breast malignancy, contain numerous genetic abnormalities.

It remains unclear whether hyperplastic breast lesions, especially with atypia, are cancer precursors or markers of increased cancer risk. Quantified comparisons of genomic alterations in coexisting lesions could address this question. Therefore, we examined allele imbalance (AI), also known as loss of heterozygosity (LOH), at 20 microsatellite markers on nine chromosome arms, in DNA from 106 samples microdissected from 17 randomly selected cancer-containing breast specimens: 13 simple (DH) and 45 atypical ductal hyperplastic (ADH) lesions, 30 in situ (DCIS) and 18 invasive ductal carcinomas (IDC). Data were analysed using regression models and generalized estimating equations. We found that AI increased as histology became more aberrant and varied with histology across the chromosome arms (p<0.0001). ADH had more AIs on 1q (p=0.03) and 16q (p=0.02) and fewer AIs on 17p (p=0.06) and 17q (p<0.0001) than on other arms. In cancers, AIs remained high on 1q and 16q, and became frequent on 17p and 17q. Concordance between AIs in ADHs and cancers exceeded the 50% expected if the lesions were separate clones in 16/20 (80%) ADHs (p=0.05), from 9/11 (82%) cases (p=0.03), and involved 41/51 (80%) evaluable markers (p=0.05). The occurrence of any AI in ADH predicted greater AI (p=0.009) and possibly lower grade (p=0.05) in coexisting cancers. Nevertheless, ADHs were not genetically identical to cancers or to each other. We found AIs discordant between ADHs and cancers (always on 1q and 16q), AIs unique to ADH (usually on 11q) and some genetic heterogeneity among coexisting ADHs. We conclude that ADH lesions are genetically advanced, with frequent alterations on 1q and 16q, and are often direct cancer precursors. Their global genetic characteristics predict features of cancers in the same breast. Nevertheless, the genetic heterogeneity detected suggests that hyperplasias and cancers may arise on a field at generalized increased cancer risk.

Increased prevalence of AI in normal-appearing epithelium is associated with breast cancer and increased breast cancer risk. The increased prevalence may reflect dysregulation, even in normal-appearing epithelium, of genomic processes contributing to cancer development. The clinical significance of genetic alterations in the subset of controls remains to be determined.