INTRODUCTION

Pleomorphic lobular carcinoma of the breast is a variant of infiltrating lobular carcinoma that has poor prognosis (1). Although it shares many of the histological features of infiltrating lobular carcinoma, the pleomorphic appearance of this variant contrasts with the cytological uniformity of classic infiltrating lobular carcinoma, thus hindering its correct identification and differentiation from ductal carcinoma (2). Identical histological features to those of infiltrating pleomorphic lobular carcinoma have recently been described in an entity termed pleomorphic lobular carcinoma in situ, which may sometimes be misinterpreted as ductal carcinoma in situ (3). Therefore, it seems that pleomorphic lobular carcinoma shares some characteristics with classic infiltrating lobular cancer, although it may also have distinctive features, thus raising the question of whether pleomorphic lobular carcinoma is a tumor of lobular nature. In this context, several studies have demonstrated that the analysis of E-cadherin expression is a powerful tool for distinguishing lobular from ductal carcinomas, even in those infiltrating and in situ tumors with indeterminate features (4, 5, 6, 7, 8).

E-cadherin is a transmembrane glycoprotein that mediates cell–cell adhesion in epithelial tissues (9). Dysfunction of the E-cadherin/catenin adhesion complex is involved in many human cancers (10, 11). Complete loss of E-cadherin expression occurs in most invasive lobular carcinomas and lobular carcinomas in situ, but not in invasive ductal cancer or ductal carcinoma in situ (4, 5, 12). Moreover, abnormal cytoplasmic expression of E-cadherin only occurs in lobular neoplasms (in a subset of tumors), whereas in about 50% of ductal tumors, reduced levels of membrane-bound E-cadherin have been reported (8, 13). Consistent with these findings, E-cadherin gene (CDH1) has been reported to be frequently mutated in infiltrating lobular carcinoma (14, 15, 16) and lobular carcinoma in situ (17), but not in ductal tumors (18). In most lobular tumors, the remaining wild-type CDH1 allele is inactivated by loss of heterozygosity (LOH) at the CDH1 locus (16q22.1) (16, 18). Genetic losses affecting the 16q region are frequently found in lobular tumors (80–100%) (16, 17, 19). Therefore, because complete loss of E-cadherin expression and high frequency of LOH at 16q22.1 are characteristic of lobular tumors (20), we have investigated whether these alterations occur in a series of 29 pleomorphic lobular breast carcinomas to confirm the lobular nature of this entity.

MATERIALS AND METHODS

Tumor Samples

This study is of a series of 29 formalin-fixed, paraffin-embedded pleomorphic lobular carcinomas that were selected according to morphological criteria (1, 21). Briefly, individual tumor cells were large and globoid, with abundant eosinophilic granular cytoplasm and irregular hyperchromatic nuclei. The tumor cells infiltrated in the classical single file and targetoid pattern of growth. Additionally, seven cases had an associated in situ component. The clinicopathological features were described in a previous study (22). Briefly, the median patient’s age was 63 years, and 71% were postmenopausal. Tumor size ranged from 1.2 to 2.5 cm. Sixteen percent of the tumors were Stage I, 50% were Stage II, 31% were Stage III, and 3% were Stage IV. Eighty-one percent of the tumors were positive for estrogen receptors, and 67%, for progesterone receptors.

Immunohistochemistry

Two-micrometer-thick tissue sections were cut onto sialinated slides, deparaffinized, and cleared and stripped of endogenous peroxidase. For heat-induced antigen retrieval, sections were immersed in boiling 10 mm sodium citrate at pH 6.5 for 2 minutes in a pressure cooker. Sections were then incubated with a primary monoclonal antibody against E-cadherin (HECD-1; Zymed, San Francisco, CA; 1:200 dilution). The slides were developed with diaminobenzidine, counterstained with hematoxylin, and subjected to qualitative and quantitative visual assessment. Normal breast tissues known to express E-cadherin, were included in each staining run as positive controls. Additionally, most studied tumor sections also included normal breast epithelium as an internal positive control. In negative controls, the primary antibodies were omitted. Composite immunoreactivity scores were obtained by adding the values of the immunoreaction intensity and the relative abundance of immunoreactive cells, as previously reported (4).

Microdissection and DNA Extraction

To avoid contamination of tumoral samples by normal cells, tumoral cells were manually microdissected from the invasive component in all tumors and from the in situ component of eight cases. Separate cell populations of normal ductal epithelium and stroma were also microdissected. DNA was extracted from 5-μm-thick sections under direct light microscopic visualization with a disposable 30-gauge needle using the previously described microdissection technique (23). Digestion of the microdissected cells was carried out following the proteinase-K-based method, as previously described (22). We consecutively analyzed microdissected tumor cells. Samples were discarded and not subjected to analysis when we were concerned about the quality of the DNA and its ability to be amplified.

Assessment of LOH at CDH1locus

Loss of heterozygosity at the CDH1 locus (16q22.1) was assessed using two highly polymorphic microsatellite markers (D16S265 and D16S752) that map close to CDH1 gene (from the Genome Database, www.gdb.org). To ensure the reliability of the results, each marker was PCR amplified twice in both the normal and tumoral DNA. Fluorescent-labeled PCR products were diluted and analyzed in an automated sequencing system (ABI Prism 3700; Applied Biosystems, PE) using Genescan software (Applied Biosystems, Warrington, UK). LOH was calculated according to a previously described formula (24).

RESULTS

Pleomorphic Lobular Carcinoma Show Complete Absence of E-Cadherin Expression

In contrast to normal ducts, where there was always strong membranous staining, all pleomorphic lobular tumors (29/29, 100%) showed complete loss of E-cadherin immunostaining (Fig. 1C). This complete absence of E-cadherin was also found in the in situ component of the tumors, implying that E-cadherin inactivation is an early event in the tumorigenesis of lobular carcinomas (Fig. 1D).

FIGURE 1
figure 1

Hematoxylin-eosin staining of an invasive pleomorphic lobular carcinoma (A) and the corresponding in situ component (B). E-cadherin expression is completely lost in pleomorphic lobular tumors (C) and its associated carcinoma in situ (D) but is retained in the membrane of epithelial cells in normal ducts. LOH at CDH1 locus is frequently detected in both invasive (E) and in situ (F) pleomorphic lobular tumors. Note that one allele is lost in the tumor (marked by arrows), in contrast to its paired normal tissue.

E-Cadherin Gene Is Inactivated by Loss of Heterozygosity in Pleomorphic Lobular Carcinomas

All pleomorphic lobular carcinomas studied were informative for at least one of the microsatellite markers used. Nineteen of 29 (65%) cases were informative for the microsatellite marker D16S752, of which 19/19 showed LOH. Using the D16S265 marker, 22/29 (76%) pleomorphic lobular carcinomas were informative, and all of these showed LOH.

The same results were obtained in the duplicate assays on all the cases analyzed.

In summary, LOH affecting CDH1 locus was detected in 27/27 (100%) informative tumors (Fig. 1E). Additionally, LOH in E-cadherin gene was also detected in all in situ components analyzed (Fig. 1F).

DISCUSSION

Pleomorphic lobular carcinoma is a distinctive subtype of invasive breast cancer and accounts for approximately 1% of all epithelial malignancies of the breast.

It is of growing interest to establish the distinctive clinicopathological features and the molecular characteristics of pleomorphic lobular carcinoma to identify this aggressive tumor clearly. Efforts have been made recently to differentiate pleomorphic lobular carcinoma from classic infiltrating lobular and ductal tumors using immunohistochemistry. For instance, all pleomorphic lobular carcinomas tumors are positive for the apocrine marker gross cystic disease fluid protein-15 (GCDFP-15), confirming that in contrast to classic infiltrating lobular cancer, pleomorphic lobular carcinoma shows apocrine differentiation (21, 25, 26). Pleomorphic lobular cancer and classic infiltrating lobular tumors share elevated levels of cytokeratin and epithelial membrane antigen, but pleomorphic lobular cancer has a higher level of chromogranin expression (26). Moreover, pleomorphic lobular carcinoma frequently tends to maintain estrogen and progesterone immunoreactivity (22, 27), although a low level of hormone receptor expression has also been reported (25, 26). With regard to proliferation and apoptosis, pleomorphic lobular cancer has been shown to overexpress Her2 and to have p53 nuclear expression in 48–80% of cases. Moreover, pleomorphic lobular cancers are generally positive for bcl-2 and have a low apoptotic index. Thus, the frequent expression of the unfavorable markers Her-2 and p53 and the generally low level of apoptosis may contribute to the aggressive behavior of pleomorphic lobular carcinoma (22, 26, 27). Unfortunately, very little is known about the cytogenetic alterations that characterize pleomorphic lobular tumors. Using flow cytometry, a higher frequency of aneuploid tumor cells in pleomorphic lobular carcinomas than in classic infiltrating lobular carcinomas was found (28). More recently, loss of heterozygosity affecting p53, Her 2, BRCA1, and ESR loci has been reported in pleomorphic lobular tumors (22).

In the current study, we have evaluated whether pleomorphic lobular carcinomas show the same alterations in E-cadherin gene and protein expression as those described in infiltrating lobular tumors. Using immunohistochemistry, we have observed that all pleomorphic lobular tumors showed complete loss of E-cadherin expression, an alteration that is characteristic of lobular tumors. These results support previous observations in infiltrating lobular carcinomas indicating that most lobular tumors, independent of their histological subtype, show complete E-cadherin inactivation (8, 29, 30). Various mechanisms are involved in the inactivation of E-cadherin in breast carcinomas. Truncating mutations in the CDH1 gene are frequent in infiltrating lobular cancers, whereas they are rarely found in ductal tumors (14, 18). Transcriptional silencing by promoter hypermethylation has also been described in ductal (31) and lobular tumors (15). However, the most frequent alteration affecting CDH1 gene in breast tumors is the loss of heterozygosity at 16q, which occurs in about 50% of ductal (19, 32) and 80% of lobular carcinomas (16, 18). Nonetheless, in contrast to lobular tumors, LOH at 16q is not associated with decreased levels of E-cadherin in invasive ductal cancer, suggesting the presence of another putative tumor suppressor gene in 16q that could play an important role in ductal tumorigenesis (19).

In our series of pleomorphic lobular carcinomas, we found that all E-cadherin-negative tumors showed LOH at CDH1 locus, thus implying a strong association between loss of E-cadherin expression and LOH at 16q21–22. Additionally, mutations and hypermethylation affecting CDH1 have been detected by us in a number of pleomorphic lobular tumors (unpublished observations), supporting the hypothesis that the same mechanisms of E-cadherin inactivation are present in pleomorphic lobular carcinomas and classic lobular cancers. In fact, we have investigated E-cadherin immunoreactivity in the in situ component of seven cases of pleomorphic lobular carcinoma and found that all of them featured complete loss of E-cadherin expression. This finding confirms that unlike the case with ductal carcinoma in situ, almost all cases of lobular carcinoma in situ lack E-cadherin expression (4, 5, 8, 12, 13). Consistent with this, it has been demonstrated in a small number of samples that E-cadherin mutations and LOH at 16q21–22 appear in the in situ component adjacent to invasive lobular carcinomas but also in lobular carcinoma in situ without invasive carcinoma (17), thus implying that E-cadherin loss is an early event in lobular tumorigenesis. Here, we report that LOH at CDH1 locus is also a common alteration in the in situ component of pleomorphic lobular carcinoma, because it was detected in all cases analyzed. Because LOH data from lobular carcinoma in situ are currently very limited, our results suggest that LOH at 16q21–22 is a frequent alteration in lobular carcinoma in situ. Moreover, recent cytogenetic studies using the comparative genomic hybridization (CGH) technique have revealed that loss of material from 16q and 17p are the most frequent alterations in lobular carcinoma in situ (33, 34).

In summary, our data demonstrate that in terms of E-cadherin inactivation, pleomorphic lobular tumors are identical to classic infiltrating lobular carcinomas and distinct from ductal tumors. Thus, pleomorphic lobular carcinoma should be considered as a variant of lobular carcinoma of the breast, despite its aggressive behavior.