INTRODUCTION

Based on clinical evidence, it has been suggested that carcinomas of the endometrium can be subdivided into estrogen-related (Type I) and non–estrogen-related (Type II) tumors (1, 2, 3). Morphologically, these two types are most frequently represented by the endometrioid and serous variants of endometrial carcinoma, respectively. More recently, molecular genetic evidence (e.g., the distribution of PTEN, K-ras, and p53 mutations and microsatellite instability) has accumulated, suggesting that different pathobiologic pathways are involved in the evolution of these two types (4, 5).

E-cadherin is a calcium-dependent transmembranous epithelial adhesion molecule, which is linked to cytoskeletal actin filaments through α- and β- (or alternatively γ-) catenin. β-catenin, in addition to its function in this cell adhesion complex, serves as a key element in the Wnt signal transduction pathway, which has been implicated in embryogenesis and carcinogenesis (reviewed in 6, 7, 8, 9, 10, 11). Cytosolic accumulation of β-catenin leads to its complex formation with transcription factors like Tcf/Lef-1, translocation into the nucleus, and induction of transcription of responsive genes, including c-myc (12), cyclin D1 (13), and c-jun and fra-1 (14; Fig. 1). Alterations in E-cadherin expression have been linked to decreased cell–cell adhesion, metastatic potential, tumor dedifferentiation, and deep myometrial invasion in endometrial and other carcinomas (15, 16). Similarly, mutations in the β-catenin–encoding gene CTNNB1 may lead to defective cell adhesion function (17). Previous studies have reported β-catenin mutations that lead to stabilization and accumulation of the molecule in 10–30% and more of both uterine and ovarian endometrioid adenocarcinomas (18, 19, 20, 21, 22, 23, 24). Alterations in β-catenin have not been reported in serous carcinomas.

FIGURE 1
figure 1

Subcellular localization and interactions of β-catenin, E-cadherin, and other molecules. Note that β-catenin functions both as a cell adhesion molecule and as a transcription cofactor.

Full size image

To evaluate differences in the expression pattern of E-cadherin and β-catenin between the two types of endometrial carcinoma, we chose to compare similarly aggressive high-grade endometrioid adenocarcinomas with serous carcinomas.

MATERIALS AND METHODS

Seventeen cases of International Federation of Gynecology and Obstetrics (FIGO) Grade III endometrioid adenocarcinoma and 17 cases of serous carcinoma were selected from the archives of the New York Presbyterian Hospital. All cases were retrospectively reviewed by the authors, and the diagnoses were confirmed using accepted criteria (3). For demographic details, the reader is referred to previously published studies performed on the same cases (25, 26). One representative formalin-fixed, paraffin-embedded tissue section from each case was submitted for immunohistochemistry. For antigen retrieval, pressure cooking or microwave pretreatment for 15 minutes was performed. Primary antibodies were obtained from Zymed Laboratories Inc. (South San Francisco, CA; E-cadherin, clone HECD-1; working dilution, 1:600) and Transduction Laboratories (Lexington, KY; β-catenin, clone 14; working dilution, 1:400). For visualization of the antigen, the 3,3′-diaminobenzidine/peroxidase-based ChemMate kit (Ventana, Tucson, AZ) was used according to the manufacturer's instructions. All steps were carried out at room temperature.

Immunostains were evaluated independently by two authors (PWS and RAS) using a scoring protocol as described elsewhere (27). Briefly, a score ranging from 0 to 12 was determined as the product of staining intensity (on a scale from 0 to 3) and the percentage of positive cells (on a scale from 0 to 4). Scoring categories were defined with scores ranging from 0 (negative), 1 to 3 (weakly), 4 to 7 (moderately), and 8 to 12 (strongly positive). If the two observers’ scores fell into the same category, their evaluation was considered concordant. If their scores fell into different categories, the third author’s (LHE) evaluation was sought, and the disagreement settled by majority vote. For β-catenin, special consideration was given to the subcellular distribution of immunoreactivity (nuclear, membranous, or cytoplasmic).

For statistical analysis, the Student’s t test was used.

RESULTS

Nuclear expression of β-catenin was observed in 8 of 17 endometrioid adenocarcinomas but in none of 17 serous carcinomas (P = .003). Conversely, moderate or strong E-cadherin expression was seen in 7 of 17 serous carcinomas but only in 1 of 17 endometrioid adenocarcinomas (P = .02; Figs. 2 and 3; Table 1). Of the 17 endometrioid adenocarcinomas, 7 showed strong nuclear β-catenin coupled with weak E-cadherin expression, 5 showed strong non-nuclear β-catenin coupled with weak E-cadherin expression, 1 showed strong nuclear β-catenin and strong E-cadherin expression, and 4 showed weak or no β-catenin coupled with weak E-cadherin expression. Thus, there was a tendency of strong β-catenin expression being associated with weak E-cadherin expression.

FIGURE 2
figure 2

β-catenin immunostain. A, poorly differentiated (International Federation of Gynecology and Obstetrics Grade III) uterine endometrioid carcinoma: nuclear reactivity in numerous cells. B, uterine serous carcinoma: membranous reactivity is present (arrow), but nuclei are negative.

Full size image
FIGURE 3
figure 3

E-cadherin immunostain. A, poorly differentiated (International Federation of Gynecology and Obstetrics Grade III) uterine endometrioid carcinoma: focal weak expression on tumor cell membranes. Nuclei are negative. B, uterine serous carcinoma: strong membranous expression. Nuclei are negative.

Full size image
TABLE 1 Staining Patterns for β-Catenin and e-Cadherin in Endometrioid Adenocarcinoma and Serous Carcinoma
Full size table

A comparable trend was not seen in serous carcinomas: of the 17 serous carcinomas, 5 showed strong non-nuclear β-catenin and strong E-cadherin expression, 5 had strong non-nuclear β-catenin and weak E-cadherin expression, 5 had both weak non-nuclear β-catenin and weak E-cadherin expression, and 2 had weak non-nuclear β-catenin with strong E-cadherin expression.

If present in serous carcinomas, β-catenin expression was non-nuclear. E-cadherin expression was restricted to the membranes in both histologic subtypes.

DISCUSSION

We undertook an immunohistochemical analysis of E-cadherin and β-catenin expression in serous carcinomas and poorly differentiated (FIGO Grade III) endometrioid adenocarcinomas. Using this approach, we attempted to minimize a potential bias that might be introduced by comparing relatively indolent low-grade endometrioid adenocarcinomas with the generally highly aggressive serous carcinomas.

Nuclear expression of β-catenin was observed in a subset of endometrioid adenocarcinomas but not in serous carcinomas. Conversely, strong membranous E-cadherin expression was identified predominantly in serous carcinoma as compared with endometrioid adenocarcinoma. Most endometrioid adenocarcinomas showed strong β-catenin expression associated with weak E-cadherin expression. Thus, we found the expression patterns of E-cadherin and β-catenin to be strongly associated with the histologic subtype. This finding provides further support for the separation of endometrioid adenocarcinoma and serous carcinoma as different entities and for their distinct molecular genetic evolution.

Clinical, morphologic and molecular genetic evidence suggests that carcinoma of the endometrium is a heterogeneous group of diseases. Based on clinical data, two major types have been identified: Type I is estrogen related and tends to have a relatively favorable prognosis, whereas Type II is non–estrogen related and follows a more aggressive course (1). Morphologically, these two types can often be correlated with the endometrioid and serous variants of endometrial carcinoma, respectively (2). Recently, molecular genetic findings have supported the view that the two types of endometrial carcinoma emerge via different pathogenetic pathways. Notably, endometrioid adenocarcinomas show mutations in the PTEN tumor suppressor gene in up to 50% and a reported frequency of approximately 28% (5) of microsatellite instability in low-grade tumors, whereas p53 mutations tend to occur late in tumor development. In contrast, serous carcinomas harbor p53 mutations in approximately 90% early on but only rarely show PTEN mutations or microsatellite instability (4, 5).

Mutational analysis of CTNNB1 has revealed stabilizing β-catenin mutations in 10–20% of endometrioid carcinomas of the endometrium and of the ovary (18, 19, 20, 22, 24). Similar figures have been described by Ikeda et al. (23), whereas Mirabelli-Primdahl et al. (21) report a 33–50% incidence of CTNNB1 mutations in endometrial carcinomas; however, the histologic subtype of the analyzed cases was not clearly specified in these two studies. For unknown reasons, the mutations tend to occur in low-grade, low-stage tumors (18, 20, 22, 24) that lack lymph node metastases (28).

Immunohistochemically, nuclear accumulation of β-catenin has been reported in ≤38% of endometrioid tumors (18, 29). This suggests that abnormalities in other elements of the Wnt-signaling pathway may be involved in the pathogenesis of endometrioid carcinomas, the common result being an up-regulation of β-catenin. Saegusa and Okayasu (30) report an association of both β-catenin mutations and nuclear accumulation with squamous differentiation in G1 and G2 endometrioid endometrial and ovarian carcinomas.

The mutational status of CTNNB1 in serous carcinoma is unknown.

Cytoplasmic β-catenin is a key element in the Wnt signal transduction pathway, and its nuclear translocation has been linked to the induction of the c-myc proto-oncogene (12), cyclin D1 (13), and c-jun and fra1 (14), among others. Interestingly, cyclin D1 has been shown to be expressed in 48% of FIGO Grade III uterine endometrioid carcinomas, as opposed to only in 15% of uterine serous carcinomas (26). Thus, the similar percentage of cases positive for both β-catenin and cyclin D1 in high-grade endometrioid adenocarcinomas may reflect the induction of cyclin D1 by β-catenin overexpression.

In nonproliferating cells, the cytosolic β-catenin pool is strictly regulated by a phylogenetically highly conserved mechanism involving a multiprotein complex including the adenomatous polyposis coli tumor suppressor protein (APC), axin, and glycogen synthase kinase 3-β (GSK3β), leading to phosphorylation and subsequent ubiquitin-dependent degradation of β-catenin (reviewed in 6, 7, 8, 9, 10, 11). E-cadherin may contribute to the elimination of β-catenin from the cytoplasm by recruiting it into the adherens junction complex, thereby preventing its translocation into the nucleus (31, 32, 33). On the other hand, as part of the adherens junction complex, β-catenin links E-cadherin through α-catenin to the cytoskeleton (Fig. 1). Thus, malfunction of β-catenin may lead to decreased cell–cell adhesion and facilitate metastatic spread of carcinoma cells (17).

Altered E-cadherin expression has been associated with decreased cell–cell adhesion, metastatic potential, tumor dedifferentiation, and deep myometrial invasion in endometrial and other carcinomas (15, 16). Concerning the endometrioid tumors, our findings coincide with previous observations that poorly differentiated endometrial carcinomas are associated with decreased E-cadherin expression (16). Unfortunately, Sakuragi et al. (16) did not specify the histological subtype of the cases analyzed in their study. As our results show, the distinction of different subtypes of endometrial carcinoma is important, because serous tumors display a divergent expression pattern for E-cadherin.

Because both E-cadherin and β-catenin are involved in cell–cell adhesion, their differential expression pattern may also reflect the different modes of tumor dissemination in endometrioid adenocarcinomas versus serous carcinomas, the latter having a predilection for intraperitoneal dissemination.

Our results suggest that E-cadherin and β-catenin immunostains could be used along with p53 stains (5, 25) to differentiate between these two types of malignancies in diagnostically difficult situations.

In summary, we found the expression patterns of β-catenin and E-cadherin in high-grade endometrial carcinomas to be associated with the histological subtype. These findings support the concept of divergent molecular genetic pathways in different types of endometrial carcinomas. The immunohistochemical staining pattern could be useful for differentiating tumor types for diagnostic purposes.