Abstract

Tumors of the central nervous system (CNS) represent a unique, heterogeneous population of neoplasms and include both benign and malignant tumors. The present study was carried out on a total of 79 archival cases of ependymal tumors in addition to a variety of other primary CNS tumors. The study entailed the use of CD99 monoclonal antibody and epithelial membrane antigen (EMA).

It was found that all 38 ependymoma cases (classic and nonclassic) showed positive membranous and intracytoplasmic CD99 immunoreactivity. Upon comparing with other CNS tumors (41 cases), it was found that CD99 could differentiate between ependymomas and nonependymal tumors, but intensity and pattern of staining were of no consequence in determining variant type or degree of histologic aggressiveness.

In regard to EMA immunoreactivity, which was restricted to the ependymoma group, 2 patterns of staining could be detected — the intracytoplasmic dotlike pattern and the ringlike pattern — but some cases were completely negative. Thus, EMA was found to be of little value in the diagnosis of ependymoma and in the differentiation between different types and grades.

CD99 can hence be recommended for use as a good marker for differentiation between ependymal and other CNS tumors. EMA expression and pattern of distribution, on the other hand, cannot be employed to determine the type of variant or the degree of tumor aggressiveness, and hence cannot predict the behavior of ependymal neoplasms.

Introduction

Ependymoma is a tumor of the central nervous system (CNS) that may be located in the brain or spinal cord. Although it may occur over a wide age range, its peak incidence is between 10 and 15 years; both sexes are equally affected.[1-3]

Such neoplasms exhibit multiple varieties of histologic patterns. The cellularity and architecture of ependymoma vary considerably, not only from case to case, but also from area to area within the same tumor. Furthermore, the cells comprising ependymomas vary from glial to epithelial in their appearance.[4]

The histologic variants of ependymoma are classified according to the World Health Organization (WHO) grading system into the following ependymal tumor types: subependymoma and myxopapillary ependymoma (grade 1 tumors), low-grade ependymoma (grade 2 tumors), and anaplastic ependymoma (grade 3 tumors). Grade 1 tumors are clinically and pathologically well-recognized entities with favorable outcomes. On the other hand, it’s difficult to achieve a sure differential diagnosis between grade 2 and 3 ependymomas.[5-7] Rarer subtypes of ependymomas include cellular, papillary, tanacytic, giant cell, epithelioid, and clear cell variants.[8-10] The identification of additional subtypes is likely, partly on the basis of gene expression profiling results.[11]

Ependymal tumors may also be confused histologically with other nonependymal tumors, such as astrocytoma, oligodendroglioma, choroid plexus neoplasms, central neurocytomas, medulloblastomas, primitive neuroectodermal tumors (PNETs), and pituitary adenomas.[12-15] Immunohistochemical assessment in such cases was found to be important in the differential diagnosis of ependymoma from other nonependymal tumors and in differentiating between the various grades of ependymoma.[16-18]

So far, the glial fibrillary acidic protein and vimentin are often used for distinguishing ependymoma from other tumors of the CNS. Nevertheless, a panel of antibodies is usually required for an accurate final diagnosis.[19]

Epithelial membrane antigen (EMA) is another marker that has been proven to be highly selective in ependymoma and is considered a marker for the differentiated forms of these tumors.[17]

On the other hand, CD99 has been less extensively studied in ependymal neoplasia. This marker is a product of the MIC2 gene. The MIC2 gene codes for a transmembrane glycoprotein that is expressed in a variety of human tissues, varying from virtually undetectable expression in some tissues to significant overexpression in immature thymocytes, Ewing’s sarcoma, and PNETs.[20,21] Also, some studies were done to detect the role of CD99 expression in ependymoma and its use in the differential diagnosis from other nonependymal tumors.[19]

The aim in this study is to evaluate the immunoreactivity and pattern of CD99 expression in ependymal neoplasms compared with other CNS tumors and to determine the validity of using CD99 as a selective marker for ependymoma. EMA expression in ependymoma will also be evaluated specifically with regard to its role in differentiation between different grades of ependymomas (ie, behavior of ependymal tumors).

Material and Methods

Seventy-nine archival cases of primary CNS tumors (38 ependymal tumors and 41 nonependymal CNS tumors) were studied. Their paraffin blocks were collected from the archives of the Pathology Department, Faculty of Medicine, Cairo University, Cairo, Egypt, during the period from May 2004 to April 2006. The 38 cases of the ependymal tumors included 29 classic ependymomas, 3 tanycytic, 3 myxopapillary, and 3 anaplastic ependymomas. The 41 nonependymal tumors consist of 6 astrocytomas, 4 central neurocytomas, 9 choroid plexus tumors, 5 medulloblastomas, 5 oligodendrogliomas, 6 pituitary adenomas, and 6 PNETs. For statistical purposes, variants other than classic ependymomas were studied as one group (nonclassic ependymomas), and the nonependymal tumors were also studied as one group.

Age of the patient, sex, site, and size of the tumor were taken from the clinical sheet of the patient. Sections from each paraffin block were cut by a microtome at 5-micron thickness, stained with hematoxylin and eosin for histopathologic revision and for immunohistochemical study by anti-CD99 monoclonal antibody for all cases, and then stained with EMA in ependymoma cases only.

Histopathologic evaluation: Classification, grading of tumors, and the histologic variants of studied cases are reviewed according to WHO classification criteria.[7,15]

Immunohistochemical evaluation: Immunohistochemical staining was conducted by the standard labeled streptavidin-biotin method (Dako, Carpinteria, California). Immunohistochemistry for CD99 was conducted with mAb clone 12E7 (DakoCytomation, n = 35) with dilution 1:20, and EMA was conducted with clone E29 (1:400, Dako). After deparaffinization and rehydration, sections were placed in 3% hydrogen peroxide for 20 minutes to inactivate endogenous peroxidase, and treated by microwave at 121°C in citrate buffer (10 mM, pH 6.0) for 10 minutes as an antigen retrieval method. After cooling to room temperature for 30 minutes, specimens were nonspecifically blocked by incubation with normal rabbit serum for 15 minutes at room temperature. Sections were incubated with the primary antibodies for 1 hour at room temperature. The sections were then subjected to a 3-step labeling procedure with streptavidin-biotin complex with 3,3′-diaminobenzidine as the chromogen, and the sections were faintly counterstained with hematoxylin.

The positive control tissue for CD99 was the normal tonsil, in which the basal and parabasal squamous epithelial cells showed moderate-to-strong distinct membranous/cytoplasmic immunostaining.[19] Sections from infiltrating ductal carcinoma of the breast were immunohistochemically stained with antibody against EMA as a positive control.

As a negative control, a tumor tissue section was processed, but the primary antibody was not added; instead, phosphate buffer solution was used. Negative controls are internal controls performed to evaluate kit selectivity.

In this study of CNS tumors, dark brown membrane staining indicated the presence of immunoreactivity for the CD99 cell-surface antigen. Intracytoplasmic or intercellular dot (ICD)-like expression of CD99 also was interpreted as immunoreactivity of the cells for CD99. This pattern of staining was assessed by the presence of single or multiple dots in the cytoplasm of or at the intercellular border of tumor cells in all ependymomas. These parameters were compared with those in the study done by Choi and colleagues.[19]

The semiquantitative grading was given by the tumor areas showing immunoreactivity of CD99 as follows: -, if absent; +, if more than 0% to 10% of tumor area; 2+, if more than 10% to 50% of tumor area; and 3+, if more than 50% of tumor area.[19]

The ependyma of a normal brain shows 3 patterns of EMA immunoreactivity: membrane immunoreactivity confined to the luminal surface and irregular punctuate, spherical, or ringlike intracytoplasmic immunoreactivity in the subependymal layer.[22]

In tumors, dotlike intracytoplasmic immunoreactivity or ringlike immunoreactivity of EMA was quantified by counting at least 5 nonoverlapping high-power fields (HPFs, 400x magnification, area of visual field). EMA staining is considered positive in the presence of at least 5 dots per HPF. The ringlike pattern appears as EMA-stained paranuclear vacuoles.

Semiquantitative grading of EMA dotlike and ringlike patterns of staining was done as follows: 0, if absent; 1+, if more than 0% to 10% of tumor area was positive; 2+, if 10% to 50% of tumor area was positively stained; and 3+, if more than 50% was stained.[17]

Statistical Analysis

Data were statistically described in terms of frequencies (number of cases) and relative frequencies (percentages). A chi-square test was performed to compare the different study variables between the study groups. Yates’ correction and Fisher’s exact test were used only when the expected frequency was found to be less than 5. A probability value (P value) less than .05 was considered statistically significant.

All statistical calculations were done with computer programs Microsoft Excel version 7 (Microsoft Corporation, New York, NY) and SPSS (Statistical Package for the Social Science; SPSS Inc., Chicago, Illinois) statistical software. Accuracy of the different diagnostic markers in the present study was represented by sensitivity, specificity, and overall accuracy.

Results

This work included 79 cases of primary CNS tumors )38 ependymal tumors and 41 nonependymal CNS tumors). Of the 29 classic ependymomas studied, 24 (82.76%) cases were intracranially located, whereas 14 of them (58.3%) were in the posterior fossa and 5 (17.24%) were located within the spinal region. Spinal location was detected in 1 tanycytic ependymoma case (1 of 3), 1 anaplastic ependymoma case (1 of 3), and in all myxopapillary ependymomas (3 of 3).

All ependymoma (classic and nonclassic) cases showed positive membranous and intracytoplasmic CD99 immunoreactivity regardless of their site ( Table 3 , Table 4 and Table 5 ). On comparing the CD99 staining pattern of classic ependymoma and nonclassic ependymomas, it was observed that the CD99 expression was moderate to strong and mostly membranous in distribution in both groups (Figures 1-4). Intracytoplasmic pattern of staining was mostly weak to moderate in both types of ependymomas. Strong staining was only detected in 3 (3 of 29) cases of classic ependymomas. These results were statistically insignificant (P = .57), although CD99 was positive in all cases of ependymomas, but it could not differentiate between classic and nonclassic variants.

Strong intracytoplasmic CD99 immunoreactivity — when found — may be suggestive of classic ependymomas because strong staining in nonclassic cases was completely absent.  

In regard to the membranous staining for CD99, there was highly statistically significant differentiation between ependymal and nonependymal tumors (P = .00). This pattern was a feature of the classic ependymal tumors and not the nonependymal ones that expressed CD99 as weak membranous positivity in a few cases (10 of 41). CD99 in this context gave very good results as a test for differentiation between both classic and nonclassic ependymomas, and the CNS tumors with a diagnostic accuracy of 100% for classic and 96.08% for nonclassic ependymomas. Test sensitivities were 100% for classic and 77.8% for nonclassic ependymomas with a specificity of 100% for classic and 100% for nonclassic ependymomas. The presence of a cytoplasmic pattern of staining was of high statistical significance in differentiating between these 2 types of tumor groups (P = .00), because nonependymal tumors were characterized by the absence of intracytoplasmic staining except in the 2 cases of choroid plexus papillomas. In such cases, the diagnostic accuracy was 97.5%, with a sensitivity of 94.74% and specificity of 100%. The values were equally high for the intracytoplasmic pattern with a 97.5% degree of accuracy, 100% sensitivity, and 95.24% specificity.

EMA immunoreactivity was studied in ependymoma cases only and 2 patterns of staining were detected: the intracytoplasmic dotlike pattern and the ringlike pattern. In regard to classic ependymomas, both staining patterns were found in 12 of 29 cases, whereas the intracytoplasmic dotlike pattern was found solely in 9 of 29 cases. Completely negative EMA staining was detected in 8 of 29 cases (Figures 1 and 3).

In nonclassic ependymomas, distinct intracytoplasmic dotlike EMA immunoreactivity was found in 5 of 9 cases, whereas the ringlike pattern was found only in 1 case of tanycytic ependymoma. Completely negative EMA staining was detected in 4 of 9 cases. The dotlike pattern of EMA immunostaining was in favor of cerebral ependymomas with a statistically significant difference (P = .029), whereas the ringlike pattern of staining was statistically insignificant in differentiation between classic and nonclassic ependymomas (P = .204) ( Table 7 and Chart 2).
 

The ringlike pattern of EMA immunostaining showed no statistical significance upon comparing cerebral with spinal ependymomas (P = .73), and dotlike pattern of EMA immunostaining is in favor of cerebral ependymomas (statistically significant finding, P = .029). This pattern was more prominent in cerebral rather than spinal ependymomas with a 76.32% degree of accuracy, 79.31% sensitivity, and 66.67% specificity for ependymomas with cerebral location.

Discussion

Glial tumors are the most common primary tumors of the CNS.[4] Almost all types of glial tumors can recur and display some degree of malignant progression depending on the histopathologic type of the tumor, grade of malignancy, its location, patient’s age, and the extent of surgical resection.[23]

The histologic classification of gliomas has been designed to identify groups of patients with different prognoses, but the grading systems for gliomas on light microscopic examination remain controversial due to the lack of direct knowledge concerning their biological behaviors.[4]

In regard to ependymomas, most cases can be diagnosed on routine histopathologic examination alone; however, they sometimes exhibit a bewildering variety of histologic patterns. The cellularity and architecture of ependymomas vary considerably, not only from case to case but also from area to area within the same tumor. Furthermore, the cells comprising ependymomas vary from glial to epithelial in their appearance.[3] So far, glial fibrillary acidic protein and vimentin are 2 markers that are used for distinguishing ependymomas from other tumors of the CNS, especially choroid plexus tumors and astrocytic tumors. Whereas individual antibodies may serve as a useful marker in certain situations, in general, a panel of antibodies is required for an accurate final diagnosis.[24]

The present study has attempted to evaluate CD99 as a possible specific marker that could selectively differentiate ependymomas from other CNS tumors. Both CD99 and EMA were also evaluated with respect to their efficacy in differentiating between the various histologic subtypes of ependymomas.

In the present study, positive CD99 immunoreactivity was detected in all ependymomas (membranous and intracytoplasmic dotlike patterns). This finding goes with the study done by Choi and colleagues[19] in which 25 cases were studied, and all were positively stained for CD99; the study done by Stevenson and colleagues[25] in which 8 of 8 cases were positive; and the study done by Hamilton and colleagues[26] in which 2 of 2 cases were positively stained. In addition, there was a study done by Weidner and Tjoe[27] in which 6 of 9 cases were positively stained for CD99, and this was the only study that contained ependymoma cases negative for CD99 immunostaining. Optimal staining could only be obtained with the mAb clone 12E7 (DakoCytomation, n = 35), which was used in all of the above-mentioned studies and in the current study. Negative cases in the study done by Weidner and Tjoe[27] occurred with the use of different monoclonal antibodies, such as mAb clone H036-1.1 and mAb clone 0-13.

The only study that has described the patterns and grades of CD99 immunostaining in ependymoma variants and other CNS tumors was that done by Choi and colleagues.[19] Other workers have focused only on the number of positively stained cases regardless of the pattern or grade of staining. In this study, moderate-to-strong membranous CD99 expression was a feature of classic ependymomas. On comparing the CD99 staining pattern of classic ependymomas with nonclassic ependymomas, it was observed that the CD99 expression was moderate to strong and mostly membranous in distribution in both groups. However, weak membranous staining was detected only in 2 nonclassic cases ( 1 case of tanycytic and 1 case of anaplastic ependymomas). These findings go with the study done by Choi and colleagues,[19] which found that the staining pattern of CD99 was not related to the histologic type of ependymoma after studying 88 cases of CNS tumors; 25 of them were diagnosed as ependymomas (13 classic, 4 tanycytic, 4 myxopapillary, and 4 anaplastic). In their study, all ependymoma cases were positive for CD99, in which the pattern of staining was predominantly membranous. Of the 25 ependymomas that they studied, 23 (92%) showed moderate-to-strong membrane staining (grades 2 and 3) and 2 cases (8%) revealed grade 1 or weak immunostaining; one of them was tanycytic; and the other was an anaplastic ependymoma, just as the varieties found in the present work.

The intracytoplasmic pattern of staining in this work was weak in 10 of 29 cases of classic ependymomas and 4 of 9 cases of nonclassic ependymomas (2 tanycytic, 1 myxopapillary, and 1 anaplastic ependymoma), whereas moderate staining was detected in 16 of 29 cases of classic and 5 of 9 cases of nonclassic ependymomas (1 tanycytic, 2 myxopapillary, and 2 anaplastic ependymomas). Strong staining was only detected in 3 of 29 cases of classic ependymomas, with no strongly stained nonclassic cases. Thus, if we find strong intracytoplasmic staining, this may be in favor of a classic ependymoma. These findings differ from those in the study done by Choi and colleagues,[19] who found strong intracytoplasmic staining in 1 of 4 tanycytic, 2 of 4 myxopapillary, and 2 of 4 anaplastic ependymomas, in addition to 5 of 13 strongly stained classic cases.

One of the important findings in this work is that although CD99 is positive in all cases of ependymoma, it cannot be used to differentiate between classic and nonclassic variants (P = .57). The study done by Choi and colleagues[19] has reached the same conclusion and found that there is no role of pattern or degree of CD99 staining in differentiation between different ependymoma variants so far. Because these studies have been done on a semiquantitative basis, more studies on a larger number of cases are needed to detect the importance of degree and pattern of CD99 immunostaining in differentiation between different ependymoma variants.

In regard to nonependymal CNS tumors, 41 cases were stained with CD99, and they were studied as one group for statistical reasons. It was found that most nonependymal CNS tumors were CD99-nonreactive (31 of 41 cases). The membranous pattern of staining was found in 10 of 41 cases, and the intracytoplasmic pattern was found in 2 of 41 cases of choroid plexus papillomas. The few positive cases of nonependymal CNS tumors were only of grade 1, ie, weakly stained.

After comparing these results with those of classic ependymomas and its variants, it was found that both the membranous and intracytoplasmic patterns of staining were of high statistical significance (P = .00) in differentiation between classic ependymomas in addition to its variants and other nonependymal tumors.

Choi and colleagues[19] studied 63 cases of nonependymal CNS tumors; 11 (17.5%) cases revealed membranous immunostaining (3 choroid plexus papillomas, 2 choroid plexus carcinomas, 1 pituitary adenoma and 1 oligodendroglioma, and 4 PNETs); strong membranous immunostaining was found only in 4 of 10 central PNET cases; and the other positively stained cases revealed weak-to-moderate immunostaining. These findings were not in accordance to those in the current study, in which nonependymal CNS tumors failed to express strong immunoreactivity in any of the studied cases. This may be attributed to the smaller number of nonependymal tumors studied, which was much lower than the other study.

In addition to moderate membranous immunostaining, the intracytoplasmic pattern was found in one case of choroid plexus papilloma (1 of 8) in the study done by Choi and colleagues,[19] but it was sparse and localized to the epithelial surface. This latter finding was similar to that of the current study in which 2 of 9 cases of choroid plexus tumors revealed weak membranous and intracytoplasmic patterns of staining. In both studies, the localization of ICD in choroid plexus tumors was different from that of ependymoma in the former; ICD was found near the epithelial surface, whereas in the latter it was randomly distributed.

Although both ependymal and choroid plexus tumors in this study shared the same feature of intracytoplasmic and membranous staining with CD99, the pattern and distribution of CD99 may be helpful to differentiate them. This is an important finding because choroid plexus papilloma is a benign nonependymal tumor that may be misdiagnosed as ependymoma on routine histopathologic examination and is also positively stained for CD99, but it is the pattern of staining and not the intensity of staining that is the crucial differentiating factor because it is completely different from that of ependymoma.

Other studies have discussed the expression of CD99 in ependymal and nonependymal CNS tumors as well, but they didn’t focus on the pattern or degree of staining, and thus they had little value in differentiation between ependymal and nonependymal CNS tumors. CD99 was found to be positive in 1 of 2 cases of glioblastoma multiforme in the study done by Ambros and colleagues,[21] in addition to 4 cases (4 of 4) of astrocytoma and 5 cases of oligodendroglioma (5 of 5) in the study by Stevenson and colleagues.[25] Astrocytoma was seen to be CD99-positive in 4 of 7 cases in the study done by Figarella-Branger and colleagues.[28]

EMA, a highly glycosylated transmembrane protein, is a diagnostic marker for epithelial differentiation also expressed by normal ependymal cells.[22,28]

In this study, the expression of EMA immunoreactivity was studied in different variants of ependymoma. Two patterns of staining were detected: the intracytoplasmic dotlike pattern, which is considered positive with the appearance of at least 5 dots per HPF after counting 5 nonoverlapping HPFs, and the ringlike pattern, which appears as intracytoplasmic lumina or vacuoles. These 2 patterns of expression were also found in ependymomas in the studies performed by Takeuchi and colleagues,[29] Vege and colleagues,[15] Hirato and colleagues,[30] Ho,[31] Cruz-Sanchez and colleagues,[3] Figarella-Branger and colleagues,[28] and Kaneko and colleagues.[24]

Of the 29 classic ependymoma cases examined in this study, 21 cases revealed a positive intracytoplasmic dotlike pattern and 12 of them revealed a ringlike pattern of immunoreactivity. All positive cases were weakly stained. Eight cases were negative. In regard to nonclassic ependymomas (collectively studied as 1 group), an intracytoplasmic dotlike pattern was found in 5 of 9 cases ( 2 tanycytic, 2 anaplastic, and 1 myxopapillary ependymomas), whereas the ringlike pattern was found only in 1 case of tanycytic ependymoma. Four out of 9 nonclassic cases were negative for EMA. These results when statistically analyzed were of negative value, indicating that both patterns of EMA staining were nonsignificant in differentiating between ependymoma grades and variants. However, a larger number of nonclassic cases is needed in additional studies to prove this finding and to make a more relevant comparison.

The above results were close to those found by Hasselblatt and Paulus,[17] who studied the pattern and extent of EMA expression in 54 ependymomas (33 classic, 2 tanycytic, 6 myxopapillary, and 13 anaplastic ependymomas). Distinct punctate intracytoplasmic EMA immunoreactivity was observed in 48 of 54 ependymomas (80%), whereas ringlike EMA staining was observed in 17 of 54 (31%) cases. Apart from the absence of EMA expression in most myxopapillary ependymomas (only 1 positive case), neither staining intensity or pattern was related to tumor type. Moreover, EMA immunoreactivity did not reveal any differences between grade 2 and grade 3 ependymomas, so EMA could not differentiate between different grades of ependymoma.

Sensitivity and specificity of punctate EMA staining for the diagnosis of ependymoma compared with other glial tumors were determined by Hasselblatt and Paulus.[17] A finding of 5 punctate dots/HPFs was associated with a sensitivity of (72%) and specificity of (81%). The presence of ringlike EMA-positive structures was less sensitive (32%), but highly specific (100%). The conclusion of this study was that distinct punctate and ringlike EMA staining may serve as a sensitive and specific marker of ependymal differentiation in glial tumors, thus aiding the diagnosis of ependymomas.[17]

The correlation between tumor grade and EMA immunoreactivity was made in previous studies, and controversial results were reached. Uematsu and colleagues[22] studied EMA expression in 20 ependymomas (13 differentiated grades 1 and 2) and 7 anaplastic ependymomas; they found that EMA was expressed in 11 of 13 differentiated cases, whereas no immunoreactivity was detected in anaplastic cases. Their results indicated that EMA is highly selective for differentiated ependymomas. A similar conclusion was also reached by Vege and colleagues,[15] who concluded that EMA is more expressed in differentiated — especially cellular — ependymomas. On the other hand, EMA expression was found to be more prominent in anaplastic ependymomas in the study done by Kaneko and colleagues.[24]

In all of the above studies, monoclonal antibody E29 was employed; however, this controversy between different studies may be due to differences in methodology and techniques of staining. The value of antigen retrieval pretreatment in detecting the dotlike EMA immunoreactivity in ependymomas was carried out by Kawano and colleagues.[32] Plain EMA immunostaining showed dotlike positivity in only 6 of 29 (21%) cases, whereas antigen retrieval pretreatment increased the number of positive cases up to 26 of 29 (90%) cases. On the basis of these results, although some false-positive findings were raised by antigen retrieval pretreatment, the study authors positively recommended adoption of that technique. Also, the extent of dotlike EMA immunoreactivity may be underestimated unless several HPFs are scrutinized.[17]

EMA immunoreactivity was also correlated with the site of ependymoma regardless of its type. The dotlike intracytoplasmic pattern was found in 23 of 29 ependymomas of cerebral location and in 3 of 9 cases with spinal location. This finding was statistically significant (P = .029), ie, EMA immunoreactivity is more prevalent in cerebral ependymoma, whereas the ringlike pattern was statistically insignificant and was not correlated with the site of ependymoma.

In the study done by Hasselblatt and Paulus,[17] of the 54 ependymomas studied, 31 cases were located in the spinal region and 23 cases were intracranially located. When EMA immunoreactivity was correlated with the site of ependymoma, it was found that neither staining pattern was related to tumor site. These findings were quite different from the current study, as previously mentioned, in which the intracytoplasmic pattern was more prevalent in intracranial ependymomas. This prevalence may be related to the larger number of intracranial cases (29) in relation to spinal cases (9) examined in the current study.

Conclusion

This work emphasizes the importance of a combination of the degree of staining and the distribution patterns of CD99 as a useful marker in differentiating between ependymal and nonependymal CNS tumors, particularly the strongly stained membranous pattern. Choroid plexus papillomas were the only group of nonependymal tumors displaying an intracytoplasmic pattern of CD99 positivity, but the distribution pattern of CD99 expression was totally different from the ependymal tumors.

EMA, on the other hand, cannot be used in such a context because it did not serve as a marker for predicting the degree of tumor cell aggressiveness or even differentiation within the group of ependymoma and its variants.

Reviewed by Dr. Ramaz Mitaishvili