3594

The role of ductal hyperplasia, usual (UDH) and atypical (ADH), as precursors to breast carcinoma (Ca) is still unclear, and distinguishing ADH from UDH and Ca in situ is difficult. Therefore, we studied genomic abnormalities by fluorescence in situ hybridization (FISH) in 27 cases of hyperplasia (10 ADH) with coexisting Ca, and 17 (4 ADH) who never had Ca. Discrete areas of normal epithelium, hyperplasia, and Ca were marked on sections for FISH. These were hybridized with 12 FISH probes in 4 multi-color panels, for 1p12, centromere (cen) 8, cen 11, and cen 17 (Breast Aneusomy Probe Set, Vysis/Abbott), TOP2A (17q21.2), MYC (8q24.21), MYH11 (16p13.11), CDH1 (16q22.1), cen 16, 1p36.3, 1q25, and COX2 (1q31.1). 20-40 cells were enumerated for each probe for each marked region, and percentage of cells with gain (>2 signals) or loss (<2 signals) of each locus determined. Student’s t-test was used to compare the percentages of cells with copy number changes of each locus between histological categories. Significant gains of nearly all loci were found in Ca versus normal (p<.05 for all loci except TOP2A). Hyperplasia versus normal regions showed significant gains for 1p12 (p=.02), cen 8 (p<.001), cen 11 (p<.001), cen 17 (p<.001), TOP2A (p<.001), MYC (p<.001), & MYH11 (p=.007). Normal regions showed no significant changes, even in the presence of coexisting Ca, but hyperplasia showed gain of 1p12 (p=.02), 1q25 (p=.05), and COX2 (p=.002), and loss of cen 11 (p<.001), cen 17 (p=.008), TOP2A (p=.004), and MYC (p=.02) in the presence of coexisting Ca compared to hyperplasia without Ca. Morphologic comparison of the 6 cases with the most copy number abnormalities with 6 cases with the least, showed that all of the former contained ADH, whereas the latter had UDH alone, except for one case with ADH. ROC analysis for the presence of coexisting Ca by gene copy changes in hyperplasia predicted that greater than 80% sensitivity and specificity might be achieved with several probe combinations (e.g. 1p12 and CEP 11), and greater than 85% sensitivity and specificity with the combination of 1p12, cen 11, and MYC. Conclusions: Florid gene copy changes often occurred in hyperplasia, but not morphologically normal epithelium, in the presence of coexisting Ca. The copy changes were more prominent in ADH than UDH. These findings suggest that high risk of occult coexisting Ca might be detected by FISH analysis of ADH in benign breast biopsies. More studies on larger cohorts will be needed to verify these findings.

[Proc Amer Assoc Cancer Res, Volume 47, 2006]