Molecular profile of atypical hyperplasia of the breast

Studies examining chromosomal content by FISH, DNA content, or nuclear morphometry of ADH and ALH showed gains or losses of whole chromosomes compared to normal breast tissue, indicating aneuploidy is a prominent feature of atypical hyperplasia (Table 3). The chromosomal changes seen in atypical hyperplasias are similar to those present in breast cancer [16‐18], and are consistent with the proposal that AHs are preneoplastic lesions and part of a continuum in the steps toward breast cancer [19, 20]. Mariuzzi et al. [17] examined the nuclear chromatin pattern in AH and found drastic changes in karyometric features, with these changes similar to those seen in comedo DCIS. Others have noted a high incidence of monoclonality (51.3%) in ADH [21], and an abnormal DNA content [6], consistent with neoplastic transformation. Eriksson et al. [22] considered the DNA cytometric findings in atypical hyperplasias to strongly indicate that, already at this early stage, complex nuclear alterations have occurred. Further evidence for this relationship is found in the studies of Cummings et al. [23] who examined specimens containing both AH and DCIS and found concordance in chromosome 1 aneuploidy between these lesions. They considered these findings to support the concept that benign proliferative breast disease is a biological precursor of in situ and invasive ductal carcinoma, the early histological changes possibly indicating a field effect with further genetic changes required for the development of a malignant phenotype [23].

Aneuploidy is an important indicator of chromosomal instability [24], resulting in significant deregulation of the transcriptome [25], aneuploidy-induced stresses [26], and contributing to further progression in the carcinogenic pathway. The causes of aneuploidy in AH are not clear; however, alterations in multiple genes known to contribute to aneuploidy have been observed in AH (Tables 4, 5, 6). For example, mutations in the tumor suppressor gene APC lead to chromosome mis-segregation as a result of kinetochore–microtubule disconnection during anaphase [27]; p53 mutations result in centrosome hyperamplification leading to multiple spindle poles and mis-segregation of chromosomes [28]; loss of p16 generates supernumerary centrosomes through centriole pair splitting, resulting in aneuploidy [29]; and BRCA1 inactivation leads to microtubule instability or centrosome amplification [30]. While speculative, one might suggest these findings indicate multiple potential pathways to the aneuploidy which is an important advanced genomic change for these lesions. Aneuploidy is also a prominent feature of high-risk normal breast tissue [31], indicating that these chromosomal changes may be a marker in general for high risk for breast cancer.

The genomic progression from normal breast tissue to atypical hyperplasia is accompanied by the development of multiple structural chromosomal abnormalities in the form of loss of heterozygosity [(LOH)/allelic imbalance (AI)] identified by microsatellite markers, and the development of large-scale chromosomal gains and losses identified by comparative genomic hybridization (CGH). These are summarized in Table 4, and it can be seen that these alterations involve multiple chromosomes and a wide range of chromosomal loci. The LOH/AI changes in AH have been shown to involve all informative markers on a chromosome arm, consistent with the pattern found in breast cancer, and in contrast to normal breast tissue and ductal hyperplasia where AI involved only single markers [32]. Atypical hyperplasia lesions are considered to be monoclonal microsatellite alterations involving both length variation and allele loss [33], and which may involve multiple genes such as Myc, EGFR, CDH13, BRCA1, p53 (Table 4). Amari et al. [34] studied 23 synchronous lesions of ADH, DCIS, and invasive carcinoma. ADH tumors with LOH were always accompanied by LOH in DCIS and IDC, consistent with ADH having a high risk of developing malignant transformation. LOH/AI is an important mechanism for loss of tumor suppressor and other genes and would contribute to the development and progression of AH.

AH has been studied by comparative genomic hybridization (CGH) to identify larger-scale structural chromosomal abnormalities. The majority of these studies utilized metaphase spreads with a resolution of ≥10 Mb. These studies identified multiple deletions and amplifications both within and between chromosomes of AH (Table 4). Chromosomes 1q, 6q, 8q, 11q, 14q were among the most frequently involved and often exhibited chromosomal gains. Gao et al. [7] reported high-level amplifications at multiple chromosomal sites, and many of these were gains, including those at 1q, 5p, 8q, 12q, 20q, and Xq, which were shared by ADH, DCIS, and invasive carcinoma. Gong et al. [35] on the other hand found five of the nine ADH lesions showing chromosome copy number alterations, with 16q loss and 17p loss being the most frequent changes. Candidate genes that might be associated with some of these losses included E-cadherin on 16q and p53 on 17p. The loss of material from chromosomes 16 and 17 was also consistent with the LOH analyses of AH (Table 4). Together, the presence of large-scale amplifications and deletions is consistent with gross chromosomal rearrangements in atypical hyperplasias. These genomically advanced changes also occur with the transition of normal adjacent breast tissue (a high-risk tissue) to breast cancer, and with the progression of HMEC’s to telomere-based crises, and are considered to be the types of chromosomal abnormalities seen in the earliest lesions of breast cancer [31, 36]. It is also noteworthy that DNA double-strand breaks (DSB) are considered to be an important etiologic mechanism in the development of both LOH [37] and GCR [38]. DSBs are an important consequence of estrogen exposure, and the average age of women with AH is over 55 years of age [4, 39], representing forty or more years (from menarche) of estrogen exposure, with the potential to induce these changes.

DNA methylation of CpG islands in the promoter region of a gene is an important mechanism for the silencing of tumor suppressor and other genes. Studies of atypical hyperplasia have identified multiple genes which may be methylated in these lesions, including tumor suppressor genes known to be important in breast carcinogenesis (Table 5; APC, BRCA1, CDH1, 14-3-3σ, HIN-1, P16, RASSF1A, RARβ). A number of cellular processes may be altered by inactivation of these genes, including cell cycle control, DNA repair, cell–cell adhesion, cell proliferation, apoptosis, cellular differentiation, and centrosome and mitotic events. Importantly, methylation in AH involves genes which are also instrumental in five of the six capabilities a cell has to acquire to become malignant (tissue invasion and metastases, limitless replicative potential, self-sufficiency in growth signals, insensitivity to antigrowth signals, and evading apoptosis) [40, 41], supporting an early and important role in breast carcinogenesis. A review of Table 5 indicates there is clearly heterogeneity in the incidence of methylation in these lesions, and in some cases methylation may also be monoallelic (CDH1) [42], further contributing to a range of gene inactivation. DNA methylation of tumor suppressor and other genes appears to play an important role in the progression of AH to breast cancer. Park et al. [43] examined methylation patterns in synchronous ADH, DCIS, and invasive ductal carcinoma. They found overall methylation levels and frequencies of APC, DLEC1, HOXA1, and RASSF1A promoter CpG islands were significantly higher in ADH than in normal breast tissue, while GRIN2B, GSTP1, HOXA1, RARβ, RUNX3, SFRP1, and TMEFF2 showed higher methylation levels and frequencies in DCIS than in ADH. This indicated that promoter methylation changed significantly in pre-invasive lesions and suggested that CpG island methylation of tumor-related genes is an early event in breast cancer progression. Hoque et al. [8] studied synchronous ADH, DCIS, and invasive ductal carcinoma. For the genes APC, CDH1, and CTNNB1 they found methylation at two or three gene loci in 25% of ADH, 28% of DCIS, and 37% in IDC. They noted atypical ductal hyperplasia and in situ carcinoma showed similar methylation patterns, suggesting that atypical hyperplasia should be considered as a well-differentiated or simply small in situ carcinoma. DNA methylation may also contribute to chromosomal abnormalities in AH. It was seen above that APC and p16INK4a regulate centrosome duplication and chromosome segregation, and loss of these genes through DNA methylation would be expected to contribute to chromosomal instability and aneuploidy [27, 29]. Together these findings indicate that DNA methylation is not only involved in the formation of AH and contributes significantly to its genomic instability, but also plays an important role in subsequent progression to malignancy.

The development of ADH and progression to DCIS and invasive breast cancer is accompanied by the acquisition of multiple gene expression differences. This is demonstrated both by gene expression profiling studies and by multiple individual gene expression studies. Ma et al. [9] examined gene expression profiling in specimens containing invasive ductal carcinoma (IDC)/DCIS/ADH/adjacent normal tissue. They found that, as compared with the patient-matched adjacent normal epithelium, significant global alterations in gene expression occurred in ADH, and these alterations were maintained in the later stages of DCIS and IDC. All of the ADH samples demonstrated a grade I gene expression signature and clustered with the low-grade DCIS and IDC samples. These three distinct stages of breast cancer (ADH, DCIS, IDC) were thus highly similar to each other at the level of the transcriptome, supporting the idea that the distinct stages of progression are evolutionary products of the same clonal origin [9].

Studies of individual genes in atypical hyperplasia indicate altered expression in multiple genes effecting a wide range of signaling pathways and cellular functions (Table 6), further supporting the gene expression profiling differences between ADH and normal adjacent tissue described above. These genes include estrogen receptor and estrogen-related genes (ER, EZH2), cell cycle genes (cyclin A1, D1, D2), loss of tumor suppressor genes (FHIT, p16, p53, RARβ) increased mitogenic activity of growth factors and oncogenes (EGFR, Her-2/neu, myc), and increased expression of transcription factors (STAT 3,5). Together these alterations may contribute to increased estrogen responsiveness, increased cell cycle progression, development of aneuploidy, decreased apoptosis, and loss of cell–cell adhesion. Alterations in many of these genes may also be associated with increased proliferation which is confirmed by increased expression of Ki67 (Table 6), a measure of cellular proliferation. Interestingly, Ki67 has also been found to be a time-varying biomarker of risk of breast cancer in women with atypical hyperplasia [44]. There is evidence that p53 is mutated in ADH [45‐47]. The presence of a dysfunctional p53 could have widespread effects in these cells including loss of cell cycle arrest and apoptosis, altered DNA repair, and genomic instability [48]. Alterations in α-tubulin and γ-tubulin are also observed in AH [49], further disrupting chromosome segregation and contributing to the aneuploidy which is observed in atypical hyperplasia (see above, numerical chromosomal changes). Importantly, it can be seen in Table 6 that for virtually all of these genes (a) the expression in ADH is altered compared with that of normal and/or non-proliferative breast tissue, and (b) this expression difference is maintained or increased in DCIS and invasive breast cancer. This is further evidence that multiple genes regulating multiple cellular processes contribute to formation of ADH and its genomic instability and are instrumental in the progression of ADH to DCIS and invasive breast cancer.

The expression of ERα in atypical hyperplasia is high [12, 50], and in some series all of the ADH lesion expressed ER [12, 51], consistent with both a prominent sensitivity to estrogens and a clonally expanded population of cells. Estradiol is an important mitogen, and the increased ER content promotes proliferation and clonal expansion, while at the same time increasing the accumulation of mutational changes. The gene EZH2 is also increased in AH and DCIS [52], and this gene transactivates genes that are commonly targeted by estrogen and WNT signaling pathways and promotes cell cycle progression in breast cancer cells [53]. The functional ER and its role in proliferation also makes it an excellent potential target for antiestrogen prevention therapy. Efficacy of tamoxifen in the prevention of breast cancer in women with AH was demonstrated in the NSABP-P1 trial [13]. Closely related to this point, the finding that antiestrogen therapy reduces breast cancer development in the ipsilateral and contralateral breast indicates that this characteristic of AH (ER positivity) is reflected in the genomic characteristics of the remaining normal breast tissue. If other genomic characteristics of AH are also reflected in these normal tissues, then the molecular profile of AH could be important for assessing future risk and responsiveness of these normal tissues. Lastly, the expression of ERβ is decreased in atypical hyperplasia [54]. ERβ is considered to play an oncosuppressive role in breast cancer [55]. Low ERβ2 expression, combined with increased ERα expression could further promote progression along the AH-DCIS/invasive carcinoma pathway.

Women with AH are at significant risk for the development of metachronous breast cancer (MBC), and it is noteworthy that the characteristics of this event for AH are very similar to those of women with sporadic breast cancer for the development of contralateral breast cancer (CBC): (a) AH contains multiple advanced genomic changes including aneuploidy, gross chromosomal rearrangements, and DNA methylation of tumor suppressor genes, all of which are common in sporadic breast cancer. (b) Both MBC and CBC are more commonly invasive breast cancer, less commonly DCIS (Table 2) [56]. (c) The cumulative incidence of breast cancer appears to increase linearly over time in both MBC [5] and CBC [56, 57]. (d) The annual incidence of breast cancer development of MBC (estimated to be approx. 0.9%/year from Hartmann [5], Fig. 2) is very similar to 0.6–0.7%/year for CBC [56, 58]. (e) The incidence of both MBC and CBC are reduced by antiestrogens [13, 59]. Together, these findings indicate that many of the features of carcinogenesis of AH are shared by sporadic breast cancer, and have strong carcinogenic potential for the future development of metachronous breast cancer.