Diagnostic Pitfalls Related to Morular Metaplasia in Endometrioid Carcinoma: An Underestimated Issue

• Morular metaplasia (MM) may conceal the glandular architecture and the typical immunophenotype of endometrioid carcinomas

• MM should be considered when assessing a gynecological neoplasm or a neoplasm of possible gynecological origin

• An immunohistochemical panel including β-catenin, ki67, CD10, and CDX2 may allow confirming or excluding a diagnosis of MM

Normal endometrium and endometrioid lesions may commonly show different types of altered differentiation, such as squamous, mucinous, secretory, tubal, and eosinophilic changes [1]. Among these, morular metaplasia (MM) displays peculiar morphological and immunophenotypical features, which make it difficult to define its nature [1-10]. MM has been described by some authors as a variant or an immature form of squamous metaplasia [11, 12]; according to other authors, it may be a type of differentiation distinct from squamous metaplasia [1, 7, 10]. We previously highlighted that MM not uncommonly shows overt squamous features although different from conventional squamous differentiation [10]. Immunohistochemically, MM lacks the typical markers of endometrioid lesions (i.e., estrogen and progesterone receptor, cytokeratin 7), while it shows an aberrant phenotype with nuclear β-catenin accumulation, low/absent ki67 expression, and CD10, CDX2, and SATB2 positivity [1-10]. Based on its morphologic and immunophenotypical features, MM may constitute a potential pitfall for pathologist. In fact, MM in a low-grade endometrioid carcinoma may mimic a solid growth pattern, leading to an overdiagnosis of high-grade carcinoma and therefore to patient overtreatment [2]. Furthermore, when the primary site of a tumor has to be defined, the peculiar immunophenotype of MM may be misleading. In our experience, these potential pitfalls are little recognized in the routine pathology practice, especially among generalist pathologists not dedicated to gynecological pathology. Here, we present a representative case which combined the major morphological and immunohistochemical pitfalls of MM, discussing the several issues related to the presence of MM in histological specimens.

A 45-year-old woman underwent total hysterectomy with bilateral salpingo-oophorectomy and rectosigmoidectomy due to an abdominal mass of 11-cm diameter. On gross examination, the mass involved full thickness both the uterine wall and the rectosigmoid wall, involving the right ovary. On histology, the tumor showed clear features of carcinoma with a predominantly solid growth pattern and a minor glandular component, with moderate nuclear atypia; necrotic-hemorrhagic changes were present (Fig. 1a). On immunohistochemistry, the tumor was completely negative for WT1, PAX8, and estrogen receptor, making a gynecological origin unlikely; cytokeratin 7 and progesterone receptor were mostly negative, with only occasional positive areas; focal positivity for vimentin, epithelial membrane antigen and racemase, and negativity for cytokeratin 20 and alpha-fetoprotein were also observed. The tumor showed patchy positivity for p16 and a p53-wild-type pattern (see online suppl. Fig. 1; for all online suppl. material, see www.karger.com/doi/10.1159/000515491). Since the endometrial involvement was minimal and CDX2 was strongly and diffusely positive (Fig. 1b), a colorectal origin was postulated in the first place. Although ovarian mucinous neoplasms may express CDX2, such possibility was not considered because our case did not show mucinous features. Since CDX2 positivity is also described in MM of endometrioid lesions, a second immunohistochemical panel including β-catenin, CD10, and ki67 was subsequently requested in order to assess such possibility. The solid areas showed strong cytoplasmic and nuclear positivity with β-catenin, strong and diffuse positivity with CD10, and a low ki67 expression; on the other hand, the focal glandular areas mainly showed membrane β-catenin positivity and high ki67 expression (Fig. 2). These features were consistent with an endometrioid carcinoma with diffuse MM. Additional sampling, especially of the contralateral uterine adnexum, revealed several foci of endometriosis. On the account of these findings, we made a diagnosis of endometrioid carcinoma of probable ovarian origin.

The presented case highlighted how the morphological and immunohistochemical features of MM may conceal the glandular architecture and the typical immunophenotype of endometrioid carcinomas, constituting a serious pitfall for pathologists.

Morphologically, MM appear as syncytial sheets of bland cells with round, ovoidal, or spindle-shaped nuclei [1, 2]. The term “morular” derives from the fact that MM is often observed in the form of small and round morules [1]; in these cases, it is usually easy to identify. However, in malignant tumors, MM may appear in the form of large, irregular, and confluent sheets, often obliterating glandular lumina and merging the glands, conferring the appearance of a solid pattern [13] (online suppl. Fig. 2). If present, overt squamous features and ghost cell keratinization may aid in identifying MM; unfortunately, such features may be focal or even completely absent [10]. The presence and extension of a solid growth pattern in endometrioid carcinoma is a crucial factor to define FIGO grade [14]. In our experience, it is not rare that low-grade (G1-2) endometrioid carcinomas with conspicuous MM are misdiagnosed as high-grade (G3) carcinomas. This may have a serious impact on the patient management, in terms of severe overtreatment. In fact, in young women who wish to preserve their fertility, a diagnosis of high-grade carcinoma on biopsy specimen precludes the possibility of a conservative management [15]. In the intraoperative assessment of frozen sections, a misdiagnosis of high-grade carcinoma may lead to unnecessary lymphadenectomy with related complications [16]. In the case of FIGO stage I on the final hysterectomy specimen, such pitfall may lead to unnecessary adjuvant treatment [15]. Cytologically, MM has typically bland nuclei and can therefore be excluded in the case of striking nuclear atypia [2]. Moreover, MM cells generally have a wider and more eosinophilic cytoplasm (resembling immature squamous features), which might distinguish them from glandular areas [1-7, 11, 12]. However, it should be remarked that solid areas in endometrioid carcinoma not uncommonly show low-grade nuclei, while MM may show mild atypia and optically clear nuclei [1], which may be accentuated in the case of artifactual changes. Furthermore, cytological differences between MM and glandular areas may be not obvious or not easily assessable, as showed in our case. All these morphological pitfalls related to MM were present in our cases, with irregular and confluent pattern mimicking solid growth, nuclear atypia more marked than usual, and no striking cytological differences with glandular areas.

Given these issues, immunohistochemistry should be recommended when MM cannot be excluded based on morphology alone. On the account of the existing literature and our own experience, an immunohistochemical panel including β-catenin, CD10, CDX2, and ki67 can be useful [1-10]. β-catenin is a membrane protein which may accumulate into the cytoplasm and the nucleus when the Wnt pathway is activated [17]. In endometrioid carcinomas with MM, β-catenin typically shows membrane staining in the glandular component and cytoplasmic/nuclear staining in the MM component [18, 19]; the nuclear expression is thought to reflect underlying mutations in CTNNB1 (β-catenin-encoding gene), which are frequent in both endometrial and ovarian carcinomas [20, 21]. In our experience, β-catenin immunostaining is very useful to define the architecture of the lesion, highlighting MM areas. Ki67 is a nuclear proliferation index used by pathologists to define the growth fraction of a tumor [9]. Although the expression of ki67 in endometrioid carcinoma is highly variable, it is typically consistent with its malignant nature [2, 8-10]. By contrast, MM areas typically show a very low ki67 expression, which may differentiate them from true solid areas [2, 8-10]. In our case, the low ki67 expression in the MM areas compared to the glandular areas was evident (Fig. 2d). CD10 is a membrane protein which is typically positive in endometrial stroma and derived neoplasms but negative in endometrial epithelium and related lesions [6]. In carcinomas with MM, CD10 is typically negative in the glandular component and diffusely positive in MM areas, clearly highlighting the latter ones [6]. CDX2 is a nuclear antigen mainly used by pathologists to confirm or exclude the gastrointestinal origin of a neoplasm, although it may be positive in adenocarcinomas of other sites [5]. In endometrioid carcinomas with MM, CDX2 is typically negative or focally positive in the glandular component, whereas MM areas show a diffuse nuclear staining [4, 5, 10]. The use of CDX2 as a marker of gastrointestinal origin might be misleading in cases of carcinomas with MM, especially considering that MM lacks the typical markers of endometrioid carcinomas. In our case, the lesion involved the uterus, the right ovary and the rectosigmoid colon, requiring a differential diagnosis among these 3 sites of origin. Estrogen and progesterone receptors, typically positive in endometrioid carcinomas, were negative. Cytokeratin 7 and PAX8, which are positive in the gynecological tract, were negative. On the other hand, although cytokeratin 20 was negative, the diffuse positivity for CDX2 led to consider a colorectal origin. Another marker of intestinal origin, that is, SATB2, was not tested; however, since it is shown that SATB2 is often positive in MM [2], its use would not have changed the diagnostic orientation. On this account, if the possibility of MM had not been considered, the diagnostic hypothesis would have been wrong. In our case, a correct diagnosis could be achieved only by testing the typical makers of MM (i.e., β-catenin and CD10 in addition to CDX2) and the proliferation marker ki67.

In summary, MM is a metaplastic change commonly observed in endometrioid lesions. On histological specimens, it may morphologically mimic a solid growth pattern, leading to overdiagnosis of high-grade carcinoma with crucial consequences on the patient management. In the case of tumors of uncertain origin, the negativity for estrogen and progesterone receptors, cytokeratin 7, and sometimes PAX8, might lead to erroneously exclude a gynecological origin; on the other hand, the positivity with CDX2 and SATB2 might be misinterpreted as indicative of a gastrointestinal origin. We believe that every pathologist, even if not dedicated to gynecological pathology, should consider MM when assessing a gynecological neoplasm or a neoplasm of possible gynecological origin. An immunohistochemical panel including β-catenin, ki67, CD10, and CDX2 may allow confirming or excluding a diagnosis of MM.

Written informed consent was obtained from the patient for publication of this case report and any accompanying images. All data were anonymized.

The authors declare no conflict of interest.

No funds were received for this study.

A.T.: study conception, pathological diagnosis, interpretation of immunohistochemical data, literature research, photomicrographs selection, and manuscript preparation; A.R.: study conception, clinical data collection, interpretation of immunohistochemical data, and manuscript preparation; A.G.: gross handling of the specimen, pathological diagnosis, interpretation of immunohistochemical data, and manuscript preparation; S.S.: literature research, histological and immunohistochemical procedures, photomicrograph selection, and manuscript preparation, F.Z.: clinical data collection, literature research, manuscript revision, and supervision; L.I.: study conception, gross handling of the specimen, pathological diagnosis, interpretation of immunohistochemical data, manuscript revision, and supervision.