Rosen's Breast Pathology, 4e

377

Ductal Carcinoma In Situ

have different genetic patterns. 235 Similar observations were reported by Marsh and Varley, 239 who analyzed LOH at 9p in multiple microdissected ducts from individual patients. LOH was found in 12 of 13 cases studied. Loss of at least one marker was detected in all subtypes (comedo, solid, cribri- form, and micropapillary), but the most extensive loss was present in comedo and cribriform lesions. The pattern of LOH in DCIS was present in the corresponding invasive car- cinoma in some but not all tumors. Aubele et al. 240 applied CGH to microdissected examples of extensive DCIS with small foci of invasive carcinoma. The procedure detected “multiple genetic changes affecting 6 to 19 different chromo- somal regions per tumor (mean 13.6 ± 5.4)” in DCIS. “Chro- mosomal alterations identified in more than one-third of the invasive lesions were mainly identical” to those in the cor- responding DCIS, except for gains of DNA on 3p and 12q, which were found more frequently in invasive carcinomas. A remarkable molecular genetic study of DCIS reported by Waldman et al. 241 employed CGH to detect chromosomal alterations in primary lesions and in subsequent local breast recurrences. Paired samples from 18 patients were studied, with all recurrences being entirely intraductal and detected 16 months to 9.3 years after initial treatment. In 17 cases, the average rate of concordance in chromosomal alterations between paired samples was 81% (range, 65% to 100%), and the pairs of lesions were morphologically similar. One pair of samples had no agreement, having 2 and 20 alterations, respectively. These findings indicate that situations classified as “recurrence” in 17 cases were instances of persistent carci- noma and that the eighteenth patient most likely had two in- dependent foci of DCIS. The mean number of CGH changes was lower in the initial lesions (8.8) than in recurrent DCIS (10.7). The degree of concordance was not ­significantly ­related to the time to recurrence. It is generally accepted that most lower grade invasive breast carcinomas evolve through a series of nonobligatory abnormal stages of preneoplastic and neoplastic growth over an uncertain period of time (most likely several years in du- ration). These stages are recognized morphologically as hy- perplasia, ADH, and DCIS, with the latter two representing the penultimate and ultimate stages of preinvasive neoplas- tic progression, respectively. Genetic analyses of matched samples of ADH, DCIS, and invasive carcinoma have identi- fied concordant allelic imbalances with similar frequencies, indicating that ADH and DCIS are nonobligate precursors to invasive carcinoma. ADH and low-grade DCIS are char- acterized by recurrent losses of 16q and 17p, and gain of 1q. The patterns of genetic aberrations in high-grade DCIS are more heterogeneous, with only rare deletions of 16q. The data, thus far, suggest that high-grade DCIS (those that are ER negative and HER positive, express “basal” markers and genomic aberrations) arise either de novo or from an as yet unidentified precursor lesion—perhaps related to apocrine lesions or to lesions like microglandular adenosis. 242 A gene signature of DCIS predictive of potential for ei- ther recurrence or invasive carcinoma has, thus far, eluded identification. 243 Livasy et al. 202 identified the “basal-like” profile (i.e., ER negative, HER2 negative, EGFR positive, and

the “nuclear grade + proliferation index” (N + P) system. The N + P methodology utilizes automated assessment of the proliferation index that has been shown to be a repro- ducible grading system for DCIS as well as for invasive car- cinoma. 224 Cases of DCIS that have a high proliferation rate by IHC, and are positive for HER2, confer a higher risk of recurrence, independent of grade and age. 225 The bcl-2 gene plays a role in control of cell growth by inhibiting apoptosis (Fig. 11.32). The expression of bcl-2 in DCIS is inversely related to grade. Loss of bcl-2 expression is most pronounced in high-grade DCIS and is directly related to p53 expression. 128 High-grade DCIS exhibit a higher rate of apoptosis (apoptotic index) than low-grade lesions. 226 Cytogenetics and Molecular Genetics Prior to the development of microdissection techniques, few genetic studies of DCIS were reported because of the technical difficulties inherent in isolating the epitheliumof these micro- scopic lesions. DNA extracted from ducts containing DCIS isolated by microdissection has provided the material for the molecular analysis of genetic alterations. Using this method, Radford et al. 227 reported loss of heterozygosity (LOH) on 17p in 29% of DCIS, compared with normal, control DNA. No significant difference in the frequency of LOH was observed among subtypes of DCIS (comedo vs. noncomedo) or in re- gard to nuclear grade. Another study revealed LOH on chro- mosome 17p13.1 in 5 of 15 informative cases studied. 228 LOH on 11p15 has been reported in a high proportion of microdis- sected samples of DCIS. 229 Chuaqui et al. 230 reported finding LOH on 11q13 in 6 (27.3%) of 22 DCIS. All of the lesions with LOH were high grade. LOH was also found in 1 of 11 lesions (9%) classified as atypical duct hyperplasia. The accumulated data suggest that LOH occurs often and early in the develop- ment of DCIS and that the sites of LOH are probable loci of one or more tumor suppressor genes. 231,232 Stratton et al. 233 detected LOH on 16q in 28% and on 17p in 29% of DCIS not associated with invasive lesions. DCIS found in conjunction with invasive carcinomas had a greater frequency of LOH on chromosomes 16q (55%) and 17p (52%). Analysis of intraductal and invasive carcinoma from the same patient has usually but not always shown similar patterns of LOH in both components. 234–237 This is of par- ticular interest because there is also a strong correlation be- tween the distribution of prognostic markers in intraductal and invasive duct carcinoma, usually with identical patterns of expression for hormone receptors, HER2, p53, EGFR, and cyclin D1. Studies by Barsky et al. 238 suggest that DCIS often has many of the biologic characteristics of invasive carcinoma, and they hypothesize that the most significant difference between in situ and invasive carcinoma lies in the inhibitory influences of myoepithelial cells. As previously noted, studies performed by these investigators indicated that one or more proteins secreted by myoepithelial cells in- hibit invasion and angiogenesis. A study of alterations in chromosome 1 revealed that the cells in individual ducts from a single specimen of DCIS may