Molecular Evidence for Progression of Microglandular Adenosis (MGA) to Invasive Carcinoma

Abstract

Microglandular adenosis (MGA) is an uncommon, benign breast lesion that is characterized by a proliferation of small uniform, round glands lined by a single layer of epithelial cells around open lumina with haphazard infiltrative growth in fibrous and fatty breast tissue. Although MGA usually has an indolent course, there is morphologic evidence that MGA can be a precursor for the development of intraductal and invasive ductal carcinoma. To investigate the possibility of such a transition, we studied 17 cases of MGA or atypical MGA some of which had given rise to carcinoma in situ (CIS) and/or invasive ductal carcinoma using the reticulin stain, immunohistochemistry (S-100, p63, Ki-67, and p53), and a molecular approach involving microdissection and high-resolution comparative genomic hybridization and MYC chromogenic in situ hybridization. MGA and carcinomas arising from MGA were typically negative for p63 and positive for S-100 and Ki-67 and occasionally positive for p53. High-resolution comparative genomic hybridization identified recurrent gains and losses in MGA (2q+, 5q−, 8q+, and 14q−) and atypical MGA (1q+, 5q−, 8q+, 14q−, and 15q−). Some examples of MGA and carcinomas arising from MGA harbored few gross chromosomal abnormalities whereas others had considerable genetic instability with widespread aberrations affecting numerous chromosomal arms. Such widespread genetic changes, together with recurrent loss of 5q and gain of 8q were reminiscent of those reported specifically for basal-like, estrogen receptor-negative, and BRCA1-associated breast tumors. Concordant genetic alterations were identified between MGA, atypical MGA, and higher risk lesions (CIS and invasive ductal carcinoma) and in some cases there was an accumulation of genetic alterations as cases “progressed” from MGA to atypical MGA, CIS, and invasive ductal carcinoma. The molecular data suggests that MGA, atypical MGA, and carcinoma arising in MGA in a single case were clonally related. This result implicates MGA as a nonobligate precursor for the development of intraductal and invasive ductal carcinoma.

Microglandular adenosis (MGA) is an uncommon, benign entity arising in the breast, which may present as an inconspicuous microscopic lesion or as a palpable mass.²¹ The term “microglandular adenosis” is believed to have been coined by McDivitt et al¹⁷ and the lesion is best known for its histologic and gross resemblance to tubular carcinoma.^3,17,21,25

MGA is a proliferation of small, uniform round glands formed by a single layer of epithelial cells around lumina containing secretion and/or calcifications. The glands haphazardly infiltrate the adipose or fibrous breast tissue.^3,17,21,25 The epithelial cells are cuboidal with clear to eosinophilic cytoplasm. Nuclei are bland, mitotic figures uncommon, and gland lumina usually have an eosinophilic secretion that can stain positively for periodic acid-Schiff or mucicarmine.²⁵ The epithelium is positive for S-100 and cytokeratins (CK) and negative for estrogen (ER) and progesterone receptors (PgR) and HER2/CerbB2.^4,8,10,14,24 The glands are surrounded by a multilayered basement membrane that stains for collagen type IV and laminin.^24,25 There is an absence of a myoepithelial cell layer around the glands of the lesion.^5,24,25

Although MGA usually has a benign clinical course, some examples of MGA have been described as giving rise to ductal carcinoma.^{8,12,14,21,22,25} MGA is rarely associated with separate, coincidental carcinoma, but in most instances carcinoma arises in and apparently from the MGA.^{8,11,12,14,21,22} A range of atypical glandular proliferations (“atypical MGA”), is also found in MGA with invasive carcinoma. Atypical MGA has features of classic MGA but greater architectural complexity and cytologic abnormality such as cellular expansion in lumina, prominent nucleoli, and presence of vesicular nuclei.^8,22 Atypical MGA probably represents an intermediary lesion in the morphologic transition from MGA to invasive carcinoma.^{8,11,12,14,22} Intraductal carcinoma arising in MGA have the propensity to retain the underlying alveolar growth pattern of the adenosis. Typically, this lesion is composed of malignant cells growing in solid nests. Basement membrane stains such as collagen type IV, laminin, or reticulin can be used to distinguish carcinoma in situ (CIS) arising in MGA from invasive carcinoma.

In contrast, invasive carcinoma that arises in MGA usually forms microscopic solid tumor masses that are appreciably larger when compared with surrounding MGA glands involved by CIS. Characteristically, invasive foci seem to be formed by coalescent growth of expanding alveolar CIS elements. Invasive carcinoma that arises in MGA is typically high grade.

MGA and atypical MGA and carcinoma arising in MGA have common immunohistochemical profiles, tending to be negative for ER, PgR, HER2, and CK5/6 and positive for CK8/18, S-100, and epidermal growth factor receptor. There is also an increasing level of positivity for Ki-67 and p53 as lesions “progress” from MGA to atypical MGA, CIS, and invasive carcinoma.^8,12,14 Until now, however, there has not been direct molecular evidence for this progression. We have investigated this by studying the molecular genetics of MGA and its morphologically atypical and malignant counterparts using chromosomal comparative genomic hybridization (CGH) to understand in more detail this apparent tumor progression.

METHODS

Cases

Archival paraffin blocks from 17 patients with a main diagnosis of MGA were retrieved from the surgical pathology files of the authors (S.J.S. and P.P.R.). The morphology of the cases was reviewed by the authors (S.J.S. and P.P.R.) to confirm the diagnosis of MGA and to classify lesions arising in the MGA.^21,25 The study was approved by the local research ethics committees.

Immunohistochemistry

Four-micron–thick formalin-fixed paraffin-embedded (FFPE) sections were used. Immunohistochemical studies were performed using the Bond-Max autostainer (Vision BioSystems, Hingham, MA). The staining was performed using a polymer-defined peroxidase detection system after antigen retrieval using epitope retrieval solution 2 (Vision BioSystems) or enzyme. Antibodies to the following antigens were used: p63 (clone 4A4, Biogenex, CA; dilution 1:100), S-100 protein (polyclonal, DakoCytomation, Carpinteria, CA; dilution 1:2000), Ki-67 (MIB-1, DakoCytomation, Carpinteria, CA; dilution 1:50), p53 (clone D07, DakoCytomation, Carpinteria, CA; dilution 1:40). Nuclear reactivity was considered positive for p63, Ki-67, and p53, whereas cytoplasmic and nuclear staining was considered positive for S-100. The percentage of cells positive in each component (MGA, atypical MGA, CIS, and invasive carcinoma] was recorded. Staining intensity was based on a 4-tier system (negative, weak, moderate, and strong). Weak staining constituted a light blush in tumor cells. Strong staining was interpreted when tumor cells demonstrated complete and dark immunoreactivity. Moderate staining was any result between what constituted weak and strong staining. Reticulin stain was also performed using standard methods.

Microdissection and DNA Extraction

Eight-micron–thick FFPE sections were cut and stained with nuclear fast red. Desired cell populations were microdissected using the PixCell Laser Capture Microdissection system (Arcturus, Mountain View, CA). Careful microdissection was performed to ensure precise capture of target lesions, as many lesions were small and to avoid contaminants, including intervening stromal cells, inflammatory cells, myoepithelial cells, and other lesion types. Estimated purity of captured target cells was 90% to 100%. DNA was extracted overnight in a humidified chamber at 55°C in 20 μL proteinase K extraction buffer. Proteinase K was inactivated at 95°C for 10 minutes.

High-resolution CGH

High-resolution CGH was used to assess whole genome copy number changes.^13,23 DNA amplification and fluorescent labeling of DNA from Laser Capture Microdissection tissue was carried out by degenerate oligonucleotide-primed polymerase chain reaction in 2 rounds and CGH was performed and analyzed using Cytovision software (Applied Imaging International, Newcastle upon Tyne, UK) according to a previously described protocol.²³ Copy number changes (gains and losses) were defined using a standard reference interval¹³ generated “in-house” to control for biases in amplification of small amounts of DNA from FFPE tissue and 99.5% confidence limits.

Chromogenic In Situ Hybridization

Chromogenic in situ hybridization (CISH) was performed using the SpotLight MYC DNA amplification probe (Zymed, South San Francisco, CA) and a previously described protocol.^15,20 Signals were evaluated by 2 of the authors (P.T.S. and L.d.S.) on a multiheaded microscope, at ×400. Thirty morphologically unequivocal neoplastic and non-neoplastic cells were counted for the presence of probe signals. Amplification was defined as >5 signals per nucleus in >50% of cancer cells, or the presence of large clusters of signal.

RESULTS

Clinicopathologic Data

The clinical and pathologic data of patients are outlined in Table 1. Of the 17 female patients (age range, 28 to 86 y; mean 57; median 59), 3 had MGA only (Fig. 1), one of whom also had a concurrent but topographically unrelated ductal CIS whereas 8 patients had invasive ductal carcinoma (7 with concurrent CIS and 1 with atypical MGA). The remaining 6 patients had atypical MGA and CIS. Thirteen of 17 (76%) patients presented with a mass, 3 patients' clinical presentation was unknown, and the remaining patient had undergone a breast reduction without clinical signs/symptoms. The right breast was involved in 11 cases; the left in 6. Two patients had a history of contralateral breast carcinoma, which was treated with mastectomy. Lymph node status was known in 3 MGA patients and all lymph nodes examined were negative (0/10, 0/10, 0/2).

TABLE 1: Clinical-Pathologic Data of Cases Studied

FIGURE 1.: Three cases of microglandular adenosis (MGA) were studied which did not have any associated lesions (“pure” MGA) (A–C). All 3 cases exhibit similarly bland morphology despite the disparate underlying genetic profiles: case 1 (A) and 2 (B) harbored no detectable gross genetic change as demonstrated by the HR-CGH karyogram (D and E, respectively), whereas case 3 (C) harbored significant genetic instability with numerous gains (green bars to the right of chromosome ideogram) and losses (red bars to the left of the chromosome ideogram) (F) (Table 3). HR-CGH indicates high-resolution comparative genomic hybridization.

Immunohistochemistry

Immunohistochemistry results were obtained in all cases (Table 2). Lack of circumferential p63 staining was found in all MGA except one that exhibited patchy staining. S-100 protein was variably present with a trend to less immunoreactivity in more morphologically advanced lesions in a given case. Ki-67 was positive in low to moderate percentage of cells in almost all cases with a mild increasing trend for a higher percentage of Ki-67 staining cells in more morphologically advanced lesions in some cases. p53 was positive (weak to moderate) in 4/9 MGA and in 6/7 atypical MGA. The reticulin stain demonstrated circumferential staining of basement membrane around foci of MGA, atypical MGA, and CIS in all cases. Stromal reticulin positivity was also seen in all cases including around foci of invasive carcinoma.

TABLE 2: Immunohistochemistry Data for S 100, p63, Ki-67, and p53 in MGA Cases

CGH Analysis of MGA and Carcinoma Arising in MGA

High resolution, chromosomal CGH data were obtained for 13 cases of MGA or atypical MGA and carcinoma arising in MGA. In total, CGH data were obtained for 36 lesions, including 12 MGA, 12 atypical MGA, 8 CIS, and 4 invasive ductal carcinomas (Table 3).

TABLE 3: Summary of DNA Copy Number Changes Identified in MGA and Associated Lesions Using Chromosomal Comparative Genomic Hybridization

Five MGA and 3 atypical MGA had no detectable genetic alterations whereas the remaining 7 MGA and 9 atypical MGA had copy number changes. Some lesions had considerable genetic instability with widespread aberrations affecting numerous chromosomal arms. The most common alterations in MGA (Fig. 2A) were gain of 2q (in 4/12 cases), loss of 5q (6/12), gain of 8q (5/12), and loss of 14q (5/12). The most common alterations in atypical MGA (Fig. 2B) were gain of 1q (4/12), loss of 5q (6/12), gain of 8q (5/12), loss of 14q (5/12), and loss of 15q (5/12).

FIGURE 2.: Summary karyogram of chromosomal alterations identified by HR-CGH in (A) MGA (n=12) and (B) atypical MGA (n=12). Black bars to the left of the chromosome ideogram represent a loss in DNA copy number at that locus in a single lesion; black bars to the right of the ideogram represent a gain in DNA copy number. HR-CGH indicates high-resolution comparative genomic hybridization; MGA, microglandular adenosis.

Three “pure” MGA were studied (cases 1 to 3; Fig. 1), 2 of which had no detectable copy number changes whereas the third had numerous gains and losses (Table 3). In the remaining cases for which CGH data was generated, there was concordance in the genomic profiles between MGA and the carcinomas arising in MGA. For example, all lesions from cases 4 and 5 had low genetic instability that correlated with absence of detectable changes in the corresponding MGA and atypical MGA. The CIS from case 5 had gain of chromosome 7 only. Cases 6, 8 to 12, 15, and 17 had numerous concordant genomic changes between the MGA and each of the carcinomas arising in MGA from the same case suggesting that there was a clonal relationship between MGA and atypical MGA, CIS, and invasive carcinoma. For example, there was a high degree of molecular association between the 4 morphologic components present in case 11, with each component harboring the following changes: 5p+, 5q−, 8q+, 9p+, 10q−, 11q+, 14q−/+, 15q−, 17q−, and Xq− (Fig. 3). Similarly, case 10 had MGA, CIS, and invasive carcinoma and each component harbored 1p−, 2p+, 5q−, 8p−, 11p−q+, 14q−, 17q−, 18q−, and Xq−. Case 8 had atypical MGA and CIS and both harbored 1p−, 1q+, 14q−, and 15q−.

FIGURE 3.: Case 11 exhibited 4 morphologic components: MGA (A), atypical MGA (C), carcinoma in situ (E), and invasive carcinoma (G). The MYC gene (8q24) copy number status, as defined by CISH, is shown for each component (B, D, F, H). (I) The summary karyogram of chromosomal alterations identified by HR-CGH demonstrates the close molecular association between all components. Black bars to the left of the chromosome ideogram represent a loss in DNA copy number at that locus and black bars to the right of the ideogram represent a gain in DNA copy number at that locus. A chromosomal region with 4 black bars to the right of the ideogram indicates that that region was gained in all 4 morphologic components. CISH indicates chromogenic in situ hybridization; HR-CGH, high-resolution comparative genomic hybridization; MGA, microglandular adenosis.

In cases 6, 10, 11, and 12, there was strong overlap in the genetic profiles from different lesions of the same case, but also evidence of accumulating genetic instability as demonstrated by increasing numbers of genetic changes as the samples evolved from MGA to invasive carcinoma. This was most notable for case 12 where the invasive carcinoma harbored all of the genetic changes detected in the atypical MGA and additional changes. Interestingly, the MGA, atypical MGA, and CIS in case 15 had overlapping copy number changes, suggesting they were closely related lesions. However, the invasive carcinoma had fewer copy number changes and only 8q+ in common with the other lesions.

CISH Analysis for MYC

Gain of chromosome 8q was a common finding in the MGA/atypical MGA, as defined by CGH, occurring in 5/12 lesions analyzed (Fig. 2). Although this is a large chromosomal arm encompassing many genes, amplification of the oncogene MYC is often associated with this gain. For validation of CGH data and to determine whether MYC was specifically gained in MGA and carcinoma arising from MGA, we assessed MYC copy number by CISH in selected cases (Table 4; Fig. 3). Increased copy number of MYC as defined by CISH correlated well with the 8q gain detected by CGH in all cases. In case 6, the MGA and associated CIS were diploid and the atypical MGA nearly triploid for MYC. By CGH, only the atypical MGA showed 8q gain involving the region of MYC. In case 11, clusters of MYC signals were detected in MGA, atypical MGA, CIS, and invasive carcinoma indicating the presence of MYC gene amplification in all components of this case (Fig. 3).

TABLE 4: Chromogenic In Situ Hybridization Data for c-MYC in MGA and Associated Lesions

DISCUSSION

MGA is a rare, infiltrative, benign lesion of the breast with an indolent clinical course. Histologic evidence of carcinoma arising from MGA has previously been documented.^{8,11,12,14,21,22} We present the first molecular investigation of carcinoma arising in MGA and report the presence of concordant, overlapping molecular alterations in MGA, atypical MGA, and carcinoma arising in MGA. These results provide strong evidence for a direct transition from MGA through atypical MGA to intraductal and invasive ductal carcinoma. In 8 of the 9 cases that had intraductal and/or invasive carcinoma arising from MGA, we were able to demonstrate concordant molecular alterations among the component lesions, indicating probable clonal evolution.

The molecular features of MGA and atypical MGA were similar to those found in basal-like carcinomas of the breast suggesting they may be related to these tumors. MGA is usually negative for ER, PgR, and Her2 and positive for S-100, epidermal growth factor receptor, and p53, characteristics shared by “triple negative” and “basal-like” carcinomas. The carcinomas that arise in MGA are frequently high grade with immunohistochemical expression similar to that of MGA.^8,12,14,24 In terms of genetic changes, there was also overlap with what we observed in MGA to that which has been documented for high-grade ER-negative, basal-like carcinomas or BRCA1-associated carcinomas. For example, there was considerable genetic instability in many of the MGA samples analyzed as demonstrated by the large number of copy number alterations and the number of chromosomes affected. The profiles were reminiscent of the “complex/saw tooth” genomic profiles described by others.^2,6,7 in which the genome was affected by numerous low-level copy number changes, p53 mutation, and with few high-level amplifications. Such molecular features are specifically associated with ER-negative, basal-like, and BRCA1-associated breast carcinomas rather than other molecular subtypes such as luminal A and B carcinomas. Furthermore, some of the more frequent alterations identified in MGA samples included loss of 5q and gain of 8q, encompassing the MYC oncogene. These findings have been specifically linked to ER-negative and high-grade invasive ductal carcinomas, basal-like carcinomas, and BRCA1-associated carcinomas.^{1,2,6,16,18,19,26}

Some examples of MGA, on the other hand, had very little genetic instability suggesting there may be molecular subgroups of MGA, which arise through different mechanisms of DNA damage. This has also been shown to occur in high-grade basal-like breast carcinomas.⁹

In summary, we demonstrate molecular genetic evidence for a transition from classic MGA through atypical MGA to CIS and invasive ductal carcinoma. These carcinomas will be treated on the basis of the most advanced component at the time of diagnosis. The challenge for the future will be to study the molecular features of “pure” MGA prospectively with clinical follow-up in an effort to identify molecular markers of risk for progression to carcinoma. The 3 examples of “pure” MGA studied here were histologically bland (Fig. 1) and 2 had no gross chromosomal abnormalities. The third had numerous copy number changes. It remains to be seen whether these cases will differ in their clinical course. However, it is likely that morphology cannot predict when there is genetic instability in MGA. Hence based on current knowledge, prospective follow-up of patients with MGA and atypical MGA will be important to develop more appropriate and evidence-based guidelines for clinical management.

ACKNOWLEDGMENT

The authors thank the Genetics Department at QML Pathology for technical resources and support.