The patient is a 44 year-old woman diagnosed with ovarian serous
carcinoma 12 years ago. She underwent unilateral salpingo-oophorectomy followed by treatment with
multi-agent chemotherapy. She subsequently suffered 2 recurrences, 5 and 7 years after diagnoses, and
has had persistent disease for the past 5 years. She has recently undergone resection of a pre-sacral
Diagnosis: Extra-axial Ependymoma
Extra-axial ependymomas are extremely uncommon tumors. They are a particularly
interesting topic to discuss not only because of their rarity, but also because of their histologic
similarity to much more commonly occurring gynecologic neoplasms. Extra-axial ependymomas have been
reported in the sacrococcygeal region, the mediastinum, omentum, lung, broad ligament and ovary.
Probably the most comprehensive descriptions have been provided in papers concerning the broad ligament
and ovarian tumors
First described by Bell, Woodruff and Scully in 1984 , broad ligament ependymoma
was characterized as a tumor bearing substantial histologic similarity to its better-known cousin
occurring in the central nervous system. The authors commented on its morphologic resemblance to
papillary serous carcinomas (and the consequent pitfalls in differential diagnosis) and described the
slowly progressive clinical course. Kleinman, Young and Scully, who also reported 3 cases of ovarian
ependymoma in 1984 , then provided a framework for understanding ependymomas in the context of other
primary ovarian neuroectodermal tumors in their report of 25 such cases . According to their
classification, primary neuroectodermal ovarian tumors could be divided into differentiated, primitive
and anaplastic tumors. Examples of differentiated tumors included ependymoma and retinal anlage tumor.
Ependymoblastoma and neuroblastoma belonged to the primitive group while glioblastoma multiforme was
typical of the anaplastic group. This publication also highlighted the correlation between clinical
outcome and morphology. The authors reiterated the frequently slowly progressive but infrequently lethal
nature of ependymoma and contrasted that with the rapidly progressive and frequently fatal course of the
primitive and anaplastic tumors.
Ovarian ependymomas are unilateral tumors, measuring between 1 and 20 cm, with a
solid and cystic gross appearance. Although it has been proposed that these tumors represent a type of
monodermal teratoma, it is very unusual to find them associated with a teratoma. They are said to be
histologically and immunophenotypically similar, if not indistinguishable, from central nervous system
teratomas, but this appears a generalization. Instead, the extra-axial ependymomas that we have studied
show a much broader range of architectural patterns and immunophenotypes than do individual typical
central nervous system tumors. Most extra-axial cases demonstrate cribriform, microcystic, macrocystic,
trabecular, pseudopapillary, papillary and solid components within each tumor. Tumors may also focally
take on the appearance of a myxopapillary ependymoma, with abundant basophilic, mucinous material
surrounding blood vessels and forming microcysts. Essentially all cases contain perivascular
pseudorosettes, but not every tumor contains ependymal rosettes. Tumor cells are cuboidal to columnar,
with pink or clear cytoplasm. Prominent cytoplasmic vacuolization can be encountered and signet ring
forms are occasionally seen. The cells constituting perivascular pseudorosettes are columnar with
characteristically thin, tapered processes that anchor tumor cells to the blood vessel. Cilia may be
prominent in cells lining cystic spaces. Nuclei are bland and round-to-oval in shape. They often have
either finely dispersed or stippled chromatin, frequently leading to the appearance of a neuroendocrine
tumor. Nuclei may contain one or more small nucleoli, but macronucleoli are not common. Nuclear grooves
can be prominent. The mitotic rate is low, perhaps less than 5 mitotic figures per 10 high power fields
in most cases.
It should not be difficult to find GFAP immunoreactivity in an extra-axial
ependymoma, but it is not uncommon to also find expression of cytokeratins and EMA. The distribution of
immunoreactivity for these different markers is geographic and appears to be related to or an expression
of tumor architecture. For example, GFAP expression is most commonly encountered in perivascular
pseudorosettes while cytokeratin expression is notable in cells lining cysts and in papillae. We have
also described (in poster 134 at this meeting) the very frequent expression of ER and PR and the
occasional expression of WT1. In addition to proposing the possible therapeutic relevance of ER/PR
expression, we showed that this finding sets apart the extra-axial tumors from those of the central
nervous system, which only rarely show ER/PR expression.
The differential diagnosis is extensive because of the great spectrum of
architectural patterns. The process of narrowing the differential diagnosis should start with a good
clinical evaluation so that metastasis to the ovary can be excluded. If the tumor has a predominant
tubular, trabecular or microacinar appearance, the differential would include endometrioid borderline
tumor or carcinoma, Brenner borderline tumor or carcinoma, adult granulosa cell tumor, Sertoli-Leydig
cell tumor, a Wolffian tumor such as a female adnexal tumor of probably Wolffian origin (FATWO), or
carcinoid. A papillary tumor would make one think about papillary serous borderline tumor or carcinoma,
endometrioid borderline tumor or carcinoma, Brenner borderline tumor or carcinoma, transitional
carcinoma, retiform Sertoli-Leydig cell tumor and mesothelioma. If the tumor is noted to be
neuroectodermal, other primary ovarian tumors that can show neuroectodermal differentiation should be
considered: mature or immature teratoma; primitive neuroectodermal tumor, Sertoli-Leydig cell tumor and
carcinosarcoma (MMMT). Attention to the nuclear features and mitotic rate should allow exclusion of
carcinoma, carcinosarcoma and primitive neuroectodermal tumor (including ependymoblastoma) in
particular. It may be difficult to exclude the other possibilities, especially if clinical information
is not forthcoming and the pathologist is not familiar with the less common entities in the differential
diagnosis. Immunohistochemistry can be useful in these cases. Aside from expression in other tumors
with neuroectodermal elements, GFAP should not be found in the tumors on the list (at least based on our
current knowledge). Of course, cytokeratin, EMA, ER/PR and even WT1 expression might confuse things a
bit, but GFAP is something you can probably hang your hat on. Many of the other tumors to consider have
characteristic immunophenotypes, something to keep in mind as you use immunohistochemistry in concert
with routine morphologic examination and clinical correlation.
Extra-axial ependymomas are largely tumors of the second, third and fourth decade.
About half of patients have had Stage I (organ-confined) disease at presentation. Approximately
one-third have had an uncomplicated clinical course, whereas the rest have suffered recurrences. These
have occurred in pelvic soft tissues, regional lymph nodes, retroperitoneum, omentum, liver and lymph
nodes of the head and neck. We are aware of only one reported death, which is remarkable given the high
rate of recurrence in this disease and that many of these patients have been followed for decades.
Surgery, chemotherapy and radiotherapy have all been tried. The use of hormonal agents is largely
unexplored, but this may prove beneficial.
- Bell DA, Woodruff JM, Scully RE. Ependymoma of the broad ligament. A report of two cases. Am J Surg Pathol. 1984 Mar;8(3):203-9.
- Kleinman GM, Young RH, Scully RE. Ependymoma of the ovary: report of three cases. Hum Pathol. 1984 Jul;15(7):632-8.
- Kleinman GM, Young RH, Scully RE. Primary neuroectodermal tumors of the ovary. A report of 25 cases. Am J Surg Pathol. 1993 Aug;17(8):764-78.