Breast cancers (BCs) are heterogeneous in their morphology, response to therapy, and clinical course.22 Molecular profiling studies have shown the existence of at least 5 different BC subtypes, each with different clinical outcomes. The luminal subtype “A” have higher levels of estrogen receptor (ER)-α and a better survival outcome compared with luminal subtypes “B.”26,46 Moreover, there is convincing genetic evidence to suggest that low-grade (LGBC) and high-grade BCs evolve through distinct evolutionary pathways.11,45 LGBCs are usually diploid/near-diploid and harbor recurrent loss of chromosome 16q and gains of chromosome 1q. In contrast, high-grade BCs are usually aneuploid with complex genetic profiles and infrequent deletion of 16q.13,23 In high-grade BCs, even when loss of 16q is present, the underlying genetic mechanism appears to be distinct from that seen in LGBCs.14 Taken together, these findings suggest that progression from LGBC to high-grade BCs is an unlikely biologic phenomenon.14,36,45
We have recently2 proposed the concept of a family of low nuclear grade breast neoplasia based on the significant coexistence of columnar cell lesions (CCLs), lobular neoplasia (LN), and atypical ductal hyperplasia/low grade ductal carcinoma in situ (ADH/low-grade DCIS) with invasive tubular carcinoma (TC), tubulolobular carcinoma (TLC), and classic invasive lobular carcinoma (ILC). In this paper, we expand our investigation to explore the phenotype of the putative precursor lesions and related cancers. We include invasive cribriform, mixed invasive tubular/classic lobular carcinoma, and invasive low nuclear grade ductal carcinoma and compare with high nuclear grade carcinoma. Phenotype was assessed using immunohistochemistry (IHC) on tissue microarrays (TMA) and full-face sections containing representative areas of the aforementioned precursor lesions, coexisting terminal ductal lobular units, and invasive carcinoma. In addition, 40 cases of precursor lesions without associated BC and 40 normal breast tissue samples from reduction mammoplasty specimens were included to test the hypothesis of whether low nuclear grade IDC, TC, invasive cribriform carcinoma (ICC), TLC, and classic ILC carcinomas share a common phenotype supporting their direct evolutionary links to CCLs, particularly flat epithelial atypia (FEA). Furthermore, we postulate a model for the biologic events that are associated with ER-α in the development of pure invasive low-grade nuclear carcinoma.
MATERIALS AND METHODS
In this study, all the available hematoxylin and eosin-stained histologic sections (range, 12 to 46 slides/case; average, 20 slides/case) of 570 successive cases of invasive low and high nuclear grade carcinoma removed by simple mastectomy were reviewed by 4 pathologists according to the recent published World Health Organisation and UK guidelines.32,48 The presence of invasive and preinvasive lesions, including CCLs, usual epithelial hyperplasia (UEH), DCIS, and LN was determined. For the purpose of this study, atypical lobular hyperplasia and lobular carcinoma in situ were grouped together under LN. A comprehensive morphologic review of CCLs was performed based on the classification system outlined by Schnitt and Vincent-Salmon.2,40 Subsequently, 4 categories were defined, including columnar cell changes (CCCs), columnar cell hyperplasia (CCH), CCCs with atypia, and CCH with atypia.2 CCC with atypia and CCH with atypia were grouped together under FEA. TMAs containing 850 of these lesions and matching terminal duct lobular units (TDLUs) were prepared (Table 1) as previously described.1
The TMAs and 40 full-face sections of benign preinvasive lesions not associated with invasive lesions, were immunohistochemically profiled for putative tumor suppressor genes (TSGs), cell cycle regulators, and proliferation and differentiation markers to immunophenotype the different lesion types (Tables 1, 2). In addition, normal breast tissues consisted of 40 biopsy specimens removed from patients (age range, 25 to 60 y; 20 cases premenopausal and 20 cases postmenopausal), who underwent reduction mammoplasty primarily because of cosmetic reasons were histologically examined and immunohistochemically profiled as well.
Immunostaining was performed as previously described.1 Positive and negative controls comprising appropriate tissue and omission of the primary antibody, respectively, were included (Table 2).
IHC Scoring and Evaluation
The IHC staining was semiquantitatively scored by 2 of the authors using both Allred and H-scores as previously described.1,4,32 Immunoreactivity was separately assessed in both epithelial and myoepithelial cells when present. The expression in tumors and putative precursor lesions was compared with those of the adjacent TDLUs. In normal TDLUs and precursor lesions, the expression of different biologic markers was quantified by estimating the percentage of positive cells as follows: 0=none, 1=<1%, 2=1% to 10%, 3=11% to 33%, 4=34% to 66%, 5=67% to 90%, 6=91% to 99%, and 7=100%.
The expression of CK19, CK7/8, CK18, CK5/6, CK14, smooth muscle actin, and vimentin was evaluated for cytoplasmic staining with a cutoff point selected at 10% determined by reference to the histogram.
The cytoplasmic staining of fragile histidine triad (FHIT) and Bcl-2 was classified as follows strong expression (H-score=210 to 300); moderate (H-score=110 to 209); mild (H-score of <110); and complete loss (P<20%). For membranous E-cadherin staining, positive expression was identified by staining in 10% or more of the cells. Positive expression was further classified into reduced (H-score ≤100) and normal (H-score >100) determined by reference to the histogram and median.
Membrane expression of Her2 was scored according to the Herceptest guidelines.32 Briefly, cases were classified as follows: negative, no membrane staining or <10% of cells staining; 1+, incomplete membrane staining in >10% of cells; 2+, >10% of cells with weak to moderate complete membrane staining; and 3+, strong and complete membrane staining in >10% of cells.32
Only nuclear reactivity was considered for ER-α and ER-β and was scored both as continuous variables (percentage of positive cells) and using the Allred score with positive expression defined as staining in >20% cells or Allred score 3 as cutoffs as previously determined.4,32
For p53 and cyclin D1, nuclear expression was categorized as follows: 0, no staining; 1, staining in <1% of cells; 2, staining in ≥1% and <10% cells; 3, staining in ≥10% and <50% cells; and 4, staining in ≥50%. p53 positivity was further subdivided into negative (score 0 or 1), borderline positive (score 2), and definite positive (scores 3 or 4).37 For cyclin D1, positive expression was defined as staining in 10% or more of the cells.37 For MIB-1, the proliferation index (PI) was determined for each case by absolute counting of positive and negative cells in each lesion (average about 500 cells/lesion) and expressed as percentage of positive cell nuclei. PI was stratified into low: <10% cells stained; intermediate: 10% to 30% cells stained; and high: ≥30% cells stained.37 For ATM, we considered lesions with <75% positive cells16 as showing reduced expression, whereas all others were classified as normal expression.
HER2 Fluorescence In Situ Hybridization
Fluorescence in situ hybridization was performed using a HER2 and a centromere 17 specific probe (Vysis, Abbott Molecular Inc, Des Plaines, IL), according to the supplier's instructions. A ratio >2.2 for HER2: centromere 17 copy number indicated amplification.
Data were analyzed using SPSS14.0 statistical software package (SPSS, Chicago, IL). The expression of different target antibodies for each lesion type was compared using Mann-Whitney U analysis or χ2 test. To assess intralesional differences for markers producing multiple outcomes, the Kruskal-Wallis test was used. The correlation between different target antibodies within each lesion type was determined using the Spearman test. The statistic significance level was set at a P value of less than 0.05. All tests were 2-tailed and a confidence interval of 95% was adopted.
The frequency of the putative precursor lesions amongst different types of breast carcinoma is summarized in Table 3. In low nuclear grade carcinoma, CCLs, ADH/low nuclear grade DCIS and LN were commonly detected (76%, 65%, and 58% of cases, respectively), whereas UEH and high nuclear grade DCIS were uncommonly seen (24% and 7.7%, respectively). Colocalization of CCL, ADH/DCIS, and/or LN and invasive low-grade nuclear carcinoma was seen in 85% patients, displaying the same cytologic-nuclear morphology in most of cases. In ILC, 90% of cases showed LN. CCL and DCIS were seen in 60% and 34% cases, respectively; and all CCLs and DCIS were associated with LN. DCIS, CCL, and LN were present altogether in the same topographic region in 34% of ILC cases.
In high-grade IDC, no preinvasive lesions were detected in 18/60 (30%) of cases, whereas DCIS, UEH, CCLs, and LN were seen in 42/60 (70%), 8/60 (13%), 6/60 (10%), and 5/60 (8%) of cases, respectively. Comedo and solid growth pattern were present in 67% and 36% of DCIS associated with high-grade IDC, whereas micropapillary and cribriform growth patterns were present in only 9% and 5%, respectively. In most cases of high-grade IDC, the DCIS lesions and invasive component displayed the same cytologic-nuclear morphology and were intimately associated with the same geographical distribution.
In the TDLUs (associated with and without invasive lesions), all luminal epithelial cells showed staining for one or more of the luminal cytokeratins, CK19, CK18, and CK8. A small proportion (<5%) of these epithelial cells coexpressed CK5/6 and/or CK14. UEH showed heterogeneous expression of luminal and basal cytokeratins (Fig. 1). Although no UEH epithelial cells showed smooth muscle actin expression, >50% of them were vimentin positive. The neoplastic cells of the FEA, ADH/low-grade DCIS lesions, and coexisting invasive LNGBC showed homogenous expression of the luminal CKs and consistently lacked expression of basal/myoepithelial markers (Tables 4–6). Thirty-three percent of high-grade IDC showed positive expression of basal cytokeratins (CK14 and/or CK5/6).
In TDLUs, 5% to 50% of luminal epithelial cells showed nuclear staining for ER-α. The expression of ER-α was significantly lower in TDLUs in premenopausal than in postmenopausal of both cancer-free (6% vs. 35%; P=0.0001) and adjacent to cancer (23% vs. 45%; P=0.0001), respectively. The expression of ER-α was significantly higher in TDLUs adjacent to preinvasive lesions with and without invasive lesions than in TDLUs of cancer-free cases in both postmenopausal and premenopause samples (P=0.001) (Tables 4, 5). ER-α expression was lower in the epithelial cells of CCC in lesions without invasion (75%) than in CCC adjacent to invasive lesions (90%). The epithelial cells of CCH, FEA, ADH/low-grade DCIS, and LN in both cancer-free and cancerous cases showed strong and diffuse positivity for ER-α (>90% positive cells). ER-α expression was significantly higher in CCC, CCH, FEA, LN and ADH/DCIS than in adjacent TDLUs in samples with and without invasive cancers. In majority of invasive low nuclear grade tumors (91%), most of the epithelial lining cells (>70%) were positively stained for ER-α.
In TDLUs of premenopausal breast reduction specimens, only a few luminal epithelial cells showed a strong Bcl-2 positivity (<10%), whereas strong Bcl-2 staining was found in 40% and >80% of luminal epithelial cells lining TDLUs of postmenopausal reduction specimens and TDLUs surrounding invasive lesions, respectively (Tables 4, 5). In addition, the epithelial cell population of CCC, CCH, FEA, LN, and ADH/low-grade DCIS, with and without invasive cancer, were almost exclusively composed of Bcl-2–positive cells and the expression was moderate to strong. Like ER-α expression, Bcl-2 expression was significantly higher in CCC, CCH, FEA, LN, and ADH/DCIS than in adjacent TDLUs in samples with and without invasive cancers (P=0.001). Sixty percent of low nuclear grade carcinomas showed positive expression of Bcl-2 (Figs. 1, 2), whereas, 69% of high nuclear grade cases showed negative expression.
A few scattered luminal cells (<1%) in premenopausal TDLUs of reduction breast specimens showed positive expression for cyclin D1 (CCND1),although the expression of cyclin D1 in TDLUs of postmenopausal reduction breast specimens, and TDLUs adjacent to preinvasive and/or invasive lesions in both cancer-free and cancerous cases was similar (<10%) (Tables 4, 5). Thirty-nine percent of CCC, 61% of CCH, 76% of FEA, 78% of ADH/low-grade DCIS, 83% of LN, and 82% of invasive LNGBC showed positive expression of cyclin D1. There was a significant stepwise increase in the percentage of cyclin D1 positive cells from normal to CCC to CCH/FEA to ADH/low-grade DCIS to invasive carcinoma (P=0.001, Kruskal-Wallis test) (Fig. 2).
There was a positive correlation between the level of ER-α expression and both Bcl-2 and cyclin D1 in the progression from normal to invasive tumors (Figs. 1, 2) (P=0.001).
In TDLUs, ER-β1 was expressed in both luminal and myoepithelial cells; >90% of luminal epithelial and myoepithelial cells expressed ER-β1. The staining pattern was identical in TDLUs of breast reduction specimens and in TDLUs adjacent to tumor. In addition, no significant difference between the expression of ER-β1 in premenopausal and postmenopausal cases was detected.5 The percentage of ER-β1–positive cells significantly decreased (P<0.001) in all putative precursor lesions (CCC, 80%; CCH, 75%; FEA, 60%; CIS, 52%) and in their coexisting invasive tumors (31%). A statistically significant (P=0.01) progressive increase in the ratio of ER-α–positive/ER-β1–positive cells was found from normal (0.3) to CCC (1.13) to CCH (1.58), to FEA/ADH/low-grade DCIS (1.7) to their coexisting invasive low-grade nuclear tumors (3.3). In invasive high nuclear grade IDC, the mean of ER-α–positive/ER-β1–positive cells ratio was 4.3.
The PI was higher in premenopausal than in postmenopausal TDLUs of reduction breast specimen (2.5% vs. 0.5%). All TDLUs, UEH, CCC, and CCH, and the majority of FEA (92%), in situ (88%), and the associated low nuclear grade carcinoma (83%) showed a low PI (<10%), whereas high-grade nuclear carcinoma displayed high PI (>10%) in more than 65% of cases. Proliferation rate was significantly elevated in CCH (5%), FEA (6.4%), ADH/DCIS (7%), and LN (4%) as compared with TDLUs (1.1%), UEH (2.5%), and CCC (1.5%) in both cancer-free and cancer samples (Tables 5, 7, 8).
Positive nuclear expression of p53 was detected in 3% and 44% of all invasive LNGBC and HNGBC, respectively, and in their coexisting in situ carcinoma, when present (Table 7). The frequency of down-regulation of BRCA1 was statistically lower in LNGBC (5%) than in HNGBC (28%) (P=0001).
ATM expression was absent or reduced in 22% and 53% of LNGBC and HNGBC, respectively. In cases showing coexistent FEA and/or in situ carcinoma with the invasive component, ATM expression was identical in both lesions (Figs. 1–3).
The epithelial cells lining the TDLUs and CCC showed strong uniform expression of FHIT protein regardless of the malignancy status. Eighty percent of low nuclear grade carcinoma showed a reduced expression of FHIT in which there was a progressive loss of FHIT in the associated putative precursor lesions (Figs. 1–4).
Neither Her2 overexpression nor HER2 gene amplification was detected in any of the precursor lesions and their associated invasive LGBC/ILC (Table 5), whereas 11% of HNGBC showed Her2 overexpression (Table 6).
All TDLUs, UEH, CCC, CCH, FEA, ADH/low-grade DCIS showed normal expression of E-cadherin, whereas LN was E-cadherin negative. The majority of TC, cribriform carcinoma, low-grade and high-grade IDC, and TLC, differed from the majority of ILC in being positive for E-cadherin (Tables 6, 7). In tubular/classic ILC mixed tumors, the separate foci of tubular and lobular carcinoma and the coexisting TDLUs, CCL, ADH/DCIS, and LN showed a remarkably similar immunoprofile apart from complete loss of E-cadherin in lobular carcinoma and LN.
Low-grade IDC, TC, ICC, classic ILC, and TLC all have a relatively favorable prognosis, possibly because of high levels of differentiation which may be the consequence of fewer genetic aberrations.45 In a recent study, we reported a high frequency of coexistence of CCLs, ADH/low-grade DCIS and LN, with TC, TLC, and ILC, respectively, suggesting that these lesions are members of a single family of low-grade precursor, in situ and invasive neoplastic lesions of the breast.2 In this report, we provide further evidence supporting this hypothesis and that CCLs are the common precursor of LNGBC. We have used an immunophenotyping approach encompassing expression of cytokeratins, cell proliferation/differentiation markers, and a number of putative TSGs.
Our results call into question the role of UEH as a BC precursor. First, it is widely accepted that UEH has a low relative risk (1.5 times) of subsequent carcinoma development.6 Although a small number of studies7,21 have demonstrated similar copy number changes in UEH and DCIS, Simpson et al44 and other studies10,27 have differentiated UEH from some forms of CCLs, and found no or only few and apparently random chromosomal changes. Furthermore, UEH does not fit well on the histologic continuum to invasive BC6 and no association between the presence of UEH and LGBC was observed2 in our earlier study. In this study, we have found that UEH showed extensive expression for CK5/6, CK14, and vimentin and heterogeneous expression for ERs, Bcl-2, FHIT, cyclin D1, and CK19/18/8. In contrast, FEA, ADH/low-grade DCIS, and LN were consistently negative for CK5/6, CK14, and vimentin and positive for ER-α, Bcl-2, cyclin D1, and CK19/18/8. The latter phenotype was shared with their colocated LGBC lesions (Fig. 1). The immunophenotype of UEH infers that this lesion either is unrelated to or at most is an orphan side branch of the LNGBC evolutionary pathway.2,45 On the basis of the morphologic, immunohistochemical, and molecular genetic data, we favor the former interpretation and believe that the most of the UEHs are not precursor lesions of LNGBC.
Presently, cancer is perceived as a clonal disease that depends on multiple genetic mutations in division-competent stem and progenitor cells.9 After transformation, these cells can become neoplastic because of deregulation of self-renewal, differentiation, membrane transport activity, telomerase activity, and antiapoptotic pathways, resulting in the ability to migrate and metastasize.9 Subsequently, the heterogeneity of BCs may derive from inherent differences in the underlying originator cell population and/or result from stochastic genetic and epigenetic events, causing different combinations of oncogene activation and loss of TSG function in normal breast stem or committed progenitor cells.9,22
Several models of stem and progenitor cells of the breast have recently been put forward.18,39 Although some suggested that the earliest progenitor cells in the epithelial bud are negative for basal cell markers and CK8/18, but positive for CK19 and Bcl-2,5 others suggested the existence of 2 distinct epithelial progenitor cell types.47 According to the latter, 1 of the precursor cells has a luminal phenotype (positive for CK8/18, CK19, MUC-1, and epithelial specific antigen; and negative for CK14 and CK5/6) and a second comprises bipotent progenitor cells (MUC-1– to±/CALLA± to+/epithelial specific antigen+) with potential to generate mixed colonies of both epithelial cells and myoepithelial cells. Our findings could be interpreted as evidence in support of the hypothesis that luminal restricted progenitor (CK8/18 and CK19) cells6 give rise to the LNGBC family or families and their precursors. However, there are several lines of evidence to suggest that the final phenotype of the tumor should not be regarded as a mere reflection of that of its progenitor cell phenotype.19,35 This is also exemplified in the mouse model developed by McCarthy and colleagues,31 where the authors inactivated Brca1 and tp53 in luminal epithelial cells of mouse mammary gland and the animals developed basal-like cancers. Furthermore, the recently described ER-α (ESR1) gene amplification24 may be a mechanism leading to the overexpression of ER.
It has been shown that estrogen can indirectly induce proliferation and inhibit apoptosis, resulting in cell proliferation and growth.12 In our study, we found that a high percentage of contiguous cells in CCLs expressed ER-α with a progressive increase in CCH, and rising even further with the appearance of atypia, in agreement with a previous study.49 In contrast, normal breast tissue showed ER-α–positive epithelial cells surrounded by ER-α–negative cells, indicating proliferation of ER-positive cells in both precancerous and cancerous breast lesions.17 In accordance with others,42,43 we have found that ER-α/ER-β1 expression ratios increased from normal toward the invasive tumors, supporting the hypothesis that ER-α and ER-β–specific pathways may have an important role in this process.38 It has been suggested that the ER-β1 gene may have tumor suppressive functions25 and there are reports demonstrating ESR2 gene silencing by gene promoter methylation. Alternatively, ER-β1 may regulate ER-α–mediated transcriptional activation, providing protection against ER-α induced hyperproliferation.12
We found the expression level of the known ER-α responsive gene cyclin D1, increased in a stepwise manner from normal to CCLs to invasive lesions.3,33 In agreement with others,8,34 we found positive and negative correlations between cyclin D1 with ER-α and ER-β1, respectively, suggesting that these receptors may oppose each other for regulating cell proliferation via cyclin D1.28 Epithelial Bcl-2 expression declined from precursor to LNGBC, as previously reported.29,50 Bcl-2 expression was associated with ER-α positivity, low MIB-1 expression, an absence of p53 mutation, and low level of Her2 expression.
We investigated the presence of a number of putative TSGs in our family of precursor and LNGBC to assess their involvement at different stages of carcinogenesis. In LGBCs, deletion of chromosome 11q23-25, containing ATM, is common.16 ATM phosphorylates p53 and BRCA1 causing their stabilization, which is required for cell cycle arrest, DNA repair, or apoptosis. A reduction in ATM protein and mRNA levels has been reported in invasive BCs.15,30 In our study, 22% of FEA, ADH/low-grade DCIS, LN, and LNGBC showed low or absent ATM expression compared with normal TDLUs. Moreover, concurrent positive expression of Bcl-2 and lack of p53 nuclear expression were found in cases with low or absent expression of ATM. In leukemia, ATM has been shown to represent an alternative regulatory mechanism of TP53 mutation20 and it is not known if it has a similar function in BC.
The function of BRCA1 is still unknown but it is proposed to be a TSG with transcriptional activity; it is involved in cell proliferation processes of mammary epithelial cells in response to hormonal stimulation, in apoptosis, control of recombination, and genome integrity after binding to proteins involved in these activities. In accordance with other studies,16,20 we found that BRCA1-associated BCs were associated with HNGBC, high PI, high frequencies of p53 alterations, and ATM loss, and negativity of estrogen.
In vivo animal studies support a tumor suppressor role for FHITgene.41 We found reduced FHIT levels in the majority of CCH, FEA, ADH/low-grade DCIS, LN and the associated invasive lesions, in accordance with others.30 Importantly, we noted an inverse correlation between FHIT and Bcl-2, cyclin D1, and ER-α expression supporting its role in the development of LNGBC.
E-cadherin (CDH1) is a TSG localized on chromosome 16q21 and is frequently lost in ILC. Our results demonstrate that although CCLs, ADH/low-grade DCIS, TC, ICC, low-grade IDC, and TLC are positive for E-cadherin, LN and ILC lack expression of this TSG and adhesion molecule. Our observations support the suggestion that loss of normal CDH1 gene expression is associated with the development of lobular differentiation.11,36,45
In summary, our findings demonstrate that FEA, ADH/low-grade DCIS, LN, and invasive low-grade BCs have remarkably similar immunophenotypes and that this phenotype is distinct from that seen in high-grade BCs. Given that the morphologic and immunohistochemical features of FEA cells are almost identical to those seen in ADH/ low-grade DCIS and LN, and that the molecular genetic changes of FEA are similar to those found in matched low-grade BCs,44 our findings suggest that FEA is a common nonobligate precursor of LGBC and ILC. Taken together, these lesions may represent a family of precursor, in situ and invasive neoplastic lesions belonging to the luminal “A” subclass of BC. Furthermore, our findings demonstrate that the balance between ER-α and ER-β expression may be important in driving cyclin D-1 and Bcl-2 expression, that p53 inactivation and Her2 overexpression/HER2 gene amplification are uncommon phenomena in this family of lesions.
1. Abd El-Rehim DM, Ball G, Pinder S, et al. High-throughput protein expression analysis using tissue microarray technology of a large well-characterised series identifies biologically distinct classes of breast cancer confirming recent cDNA expression analyses. Int J Cancer. 2005;116:340–350.
2. Abdel-Fatah TMA, Powe DG, Hodi Z, et al. High frequency of coexistence of columnar cell lesions, lobular neoplasia and low grade ductal carcinoma in situ with invasive tubular carcinoma and invasive lobular carcinoma. Am J Surg Pathol. 2007;13:417–426.
3. Alle K, Henshall S, Field A, et al. Cyclin D1 protein is overexpressed in hyperplasia and intraductal carcinoma of the breast. Clin Cancer Res. 1998;4:847–854.
4. Allred DC, Harvey JM, Berado M, et al. Prognostic and predictive factors in breast cancer by immunohistochemical analysis. Mod Pathol. 1998;11:155–168.
5. Anbazhagan R, Osin PP, Bartkova J, et al. The development of epithelial phenotypes in the human fetal and infant breast. J Pathol. 1998;184:197–206.
6. Arpino G, Laucirica R, Elledge RM. Premalignant and in situ breast disease: Biology and clinical implications. Ann Internal Med. 2005;143:446–457.
7. Aubele MM, Cummings MC, Mattis AE, et al. Accumulation of chromosomal imbalances from intraductal proliferative lesions to adjacent in situ and invasive ductal breast cancer. Diagn Mol Pathol. 2000;9:14–19.
8. Bardin A, Boulle N, Lazennec G, et al. Loss of ER beta expression as a common step in estrogen-dependent tumor progression. Endocr Relat Cancer. 2004;11:537–551.
9. Behood F, Rosen JM. Will cancer stem cells provide new therapeutic targets? Carcinogenesis. 2005;26:703–711.
10. Böcker W, Buerger H, Schmitz K, et al. Ductal epithelial proliferations of the breast: a biological continuum? Comparative genomic hybridization and high-molecular-weight cytokeratin expression patterns. J Pathol. 2001;195:415–421.
11. Buerger H, Mommers EC, Littmann R, et al. Ductal invasive G2 and G3 carcinomas of the breast are the end stages of at least two different lines of genetic evolution. J Pathol. 2001;194:165–170.
12. Cheng GJ, Li Y, Omoto Y, et al. Differential regulation of estrogen receptor (ER) alpha and ER beta in primate mammary gland. J Clin Endocrinol Metab. 2005;90:435–444.
13. Chin K, DeVaries S, Fridlyyand J. Genomic and transcriptional aberrations linked to breast cancer pathophysiologies. Cancer Cell. 2006;10:529–541.
14. Cleton-Jansen AM, Buerger H, ter Haar N, et al. Different mechanisms of chromosome 16 loss of heterozygosity in well- versus poorly differentiated ductal breast cancer. Genes Chromosom Cancer. 2004;41:109–116.
15. Cuatrecasas M, Santamaria G, Velasco M, et al. ATM gene expression is associated with differentiation and angiogenesis in infiltrating breast carcinomas. Histol Histopathol. 2006;21:149–156.
16. Ding SL, Sheu LF, Yu JC, et al. Abnormality of the DNA double-strand-break checkpoint/repair genes, ATM, BRCA1 and TP53, in breast cancer is related to tumour grade. Br J Cancer. 2004;90:1995–2001.
17. Dontu G, El-Ashry D, Wicha MS. Breast cancer, stem/progenitor cells and the estrogen receptor. Trends Endocrinol Metabol. 2004;15:193–197.
18. Dontu G, Liu SL, Wicha MS. Stem cells in mammary development and carcinogenesis-implications for prevention and treatment. Stem Cell Reviews. 2005;1:207–213.
19. Fletcher CDM. Expert Commentary
20. Gasco M, Shami S, Crook T. The p53 pathway in breast cancer. Breast Cancer Res. 2002;4:70–76.
21. Gong G, DeVries S, Chew KL, et al. Genetic changes in paired atypical and usual ductal hyperplasia of the breast by comparative genomic hybridization. Clin Cancer Res. 2001;7:2410–2414.
22. Gusterson BA, Ross DT, Heath VJ, et al. Basal cytokeratins and their relationship to the cellular origin and functional classification of breast cancer. Breast Cancer Res. 2005;7:143–148.
23. Hicks J, Krasnitz A, Lakshmi B. Novel patterns of genome rearrangement and their association survival in breast cancer. Genome Res. 2006;16:1465–1479.
24. Holst F, Stahl PR, Ruiz C, et al. Estrogen receptor alpha (ESR1) gene amplification is frequent in breast cancer. Nat Genet. 2007;39:655–660.
25. Iwao K, Miyoshi Y, Egawa C, et al. Quantitative analysis of estrogen receptor-alpha and -beta messenger RNA expression in breast carcinoma by real-time polymerase chain reaction. Cancer. 2000;89:1732–1738.
26. Jacquemier J, Ginestier C, Rougemont J, et al. Protein expression profiling identifies subclasses of breast cancer and predicts prognosis. Cancer Res. 2005;65:767–779.
27. Jones C, Merrett S, Thomas VA, et al. Comparative genomic hybridization analysis of bilateral hyperplasia of usual type of the breast. J Pathol. 2003;199:152–156.
28. Liu MM, Albanese C, Anderson CM, et al. Opposing action of estrogen receptors alpha and beta on cyclin D1 gene expression. J Biol Chem. 2002;277:24353–24360.
29. Luna-More S, Weil B, Bautista D, et al. Bcl-2 protein in normal, hyperplastic and neoplastic breast tissues. A metabolite of the putative stem-cell subpopulation of the mammary gland. Histol Histopathol. 2004;19:457–463.
30. Man SM, Ellis IO, Sibbering M, et al. High levels of allele loss at the FHIT and ATM genes in non-comedo ductal carcinoma in situ and grade 1 tubular invasive breast. J Pathol. 1997;182:A1–A1.
31. McCarthy A, Savage K, Gabriel A, et al. A mouse model of basal-like breast carcinoma with metaplastic elements. J Pathol. 2007;211:389–398.
32. NHSBSP R. Pathological Reporting of Breast Disease: A Joint Document Incorporating the Third Edition of the NHS Breast Screening Programmes Guidelines for Pathology Reporting in Breast Cancer Screening and the Second Edition of the Royal College of Pathologists' Minimum Dataset for Breast Cancer Histopathology. Sheffield: NHSBSP; 2005.
33. Oh YL, Choi JS, Song SY, et al. Expression of p21 (Waf1), p27 (Kip1) and cyclin D1 proteins in breast ductal carcinoma in situ: relation with clinicopathologic characteristics and with p53 expression and estrogen receptor status. Pathol Int. 2001;51:94–99.
34. Paruthiyil S, Parmar H, Kerekatte V, et al. Estrogen receptor beta inhibits human breast cancer cell proliferation and tumor formation by causing a G(2) cell cycle arrest. Cancer Res. 2004;64:423–428.
35. Reis-Filho JS. Re: Korsching et al. The origin of vimentin expression in invasive breast cancer: epithelial-mesenchymal transition, myoepithelial histogenesis or histogenesis from progenitor cells with bilinear differentiation potential? J Pathol. 2005;206:451–457. J Pathol. 2005;207:367–369.
36. Reis-Filho JS, Simpson PT, Gale T, et al. The molecular genetics of breast cancer: the contribution of comparative genomic hybridization. Pathol Res Pract. 2005;201:713–725.
37. Reis-Filho JS, Savage K, Lambros MBK, et al. Cyclin D1 protein overexpression and CCND1 amplification in breast carcinomas: an immunohistochemical and chromogenic in situ hybridisation analysis. Mod Pathol. 2006;19:999–1009.
38. Roger P, Sahla ME, Makela S, et al. Decreased expression of estrogen receptor beta protein in proliferative preinvasive mammary tumors. Cancer Res. 2001;61:2537–2541.
39. Russo J, Mailo D, Hu YF, et al. Breast differentiation and its implication in cancer prevention. Clin Cancer Res. 2005;11:931S–936S.
40. Schnitt SJ, Vincent-Salomon A. Columnar cell lesions of the breast. Adv Anat Pathol. 2003;10:113–24.
41. Sevignani C, Calin GA, Cesari R, et al. Restoration of fragile Histidine triad (FHIT) expression induces apoptosis and suppresses tumorigenicity in breast cancer cell lines. Cancer Res. 2003;63:1183–1187.
42. Shaaban AM, O'Niell PA, Davies MP, et al. Declining estrogen receptor beta expression defines malignant progression of human breast neoplasia. Am J Surg Pathol. 2003;27:1502–1512.
43. Shaw JA, Udokang K, Mosquera JM, et al. Oestrogen receptors alpha and beta differ in normal human breast and breast carcinomas. J Pathol. 2002;198:450–457.
44. Simpson PT, Gale T, Reis-Filho JS, et al. Columnar cell lesions of the breast: the missing link in breast cancer progression? A morphological and molecular analysis. Am J Surg Pathol. 2005;29:734–746.
45. Simpson PT, Reis-Filho JS, Gale T, et al. Molecular evolution of breast cancer. J Pathol. 2005;205:248–254.
46. Sorlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of breast carcinomas distinguish tumour subclasses with clinical implications. Proc Natl Acad Sci USA. 2001;98:10869–10874.
47. Stingl J, Eaves CJ, Zandieh I, et al. Characterization of bipotent mammary epithelial progenitor cells in normal adult human breast tissue. Breast Cancer Res Treat. 2001;67:93–109.
48. Tavassoli FA, Devilee P. Pathology and Genetics of Tumours of the Breast and Female Genital Organs. Lyon: IARS Press; 2003.
49. Tremblay G, Deschenes J, Alpert L, et al. Overexpression of estrogen receptors in columnar cell change and in unfolding breast lobules. Breast J. 2005;11:326–332.
50. Yang QF, Sakurai T, Jing XF, et al. Expression of Bcl-2, but not Bax, correlates with estrogen receptor status and tumor proliferation in invasive breast carcinoma. Pathol Int. 1999;49:775–780.