Case 4 -

Anaplastic Clear Cell Ependymoma, WHO Grade III (Modified from USCAP Short Course #37 Syllabus by Arie Perry and Daniel Brat) Arie Perry, M.D. Washington University School of Medicine St. Louis, MO, USA

Diagnostic Features: Sheets of clear cells with "fried egg" appearance, sharp demarcation, perivascular pseudorosettes, occasional signet ring-like cells, "chicken wire" capillary network, multiple foci of endothelial hyperplasia, scattered mitotic figures, focal necrosis. IHC: GFAP-positive processes highlighting perivascular pseudorosettes; native neurofilament-positive axons pushed to periphery of tumor (i.e., solid or non-infiltrative growth pattern); dot-like EMA positivity, c/w intracytoplasmic lumina; tumor negative for neuronal markers (synaptophysin, neurofilament, Neu-N) and CD99; MIB-1 LI ranging from 10% to 50%; Polysomy 1 and 19 (i.e., negative for 1p/19q deletions) and normal 2 copies of chromosome 18 by FISH. Case 4 - Figure 1 Click to view with ImageScope Click to view with a Web-Based Viewer Ependymomas Ependymomas are tumors that arise from the ependymal lining of the ventricular system in the central nervous system. The major forms recognized by the WHO are: Ependymoma, WHO grade II; Anaplastic Ependymoma, WHO grade III; Myxopapillary Ependymoma, WHO grade I; and Subependymoma, WHO grade I [1] . Additional variants include cellular, clear cell, papillary, and tanycytic ependymomas. Ependymoma, grade II and Anaplastic Ependymoma, grade III, represent a biologic spectrum of increasing malignancy [1, 2] . These are generally tumors of children and young adults that arise most commonly in the 4th ventricle, the spinal cord, and supratentorial brain parenchyma [2, 3, 4, 5] . They account for 10% of brain tumors in children and are among the most frequent brain tumors under the age of 3-yrs-old. The most common location for ependymomas in childhood is the 4th ventricle. Supratentorial tumors occur most often in the parenchyma of the cerebral hemispheres, often in a slightly older age group. Spinal ependymomas occur most often in adults (30-40 yrs) and are most frequent in the cervical and cervico-lumbar regions. The clinical symptoms of ependymomas depend on patient age and location [6, 7] . Posterior fossa tumors in infants may present with an enlarging head due to increased intracranial pressure in the setting of incompletely fused cranial sutures. In slightly older children, increased intracranial pressure and hydrocephalus are associated with headaches, nausea and vomiting, dizziness and ataxia. Spinal ependymomas present clinically with sensory and motor deficits that correspond to the spinal level involved. By neuroimaging, ependymomas are generally solid with well defined borders. They push aside adjacent nervous system rather than invading it. In the posterior fossa, ependymomas typically fill the fourth ventricle and displace the cerebellum posteriorly. It is often difficult to determine if a posterior fossa tumor arises from the cerebellum and extends into the fourth ventricle (as expected for medulloblastoma) or if tumors arise in the ventricle and extends into the cerebellum (typical of ependymoma). Ependymomas in this location may extend into the adjacent subarachnoid space by exiting through the foramina of Luschka. Ependymomas are contrast-enhancing tumors, but the degree is variable and the pattern is often heterogeneous. The histopathology of ependymoma, grade II and anaplastic ependymoma, grade III is characterized by tumor cells that form true ependymal rosettes or rosettes around central blood vessels (ependymal pseudorosettes) [1] . Pseuodorosettes consist of ependymal tumor cells oriented around a central blood vessel with long fibrillar processes that extend to the vessel and give a pinwheel appearance. True rosettes and elongated canals, form a lumen in their center. Ependymomas show a wide range of cellularity and degrees of fibrillarity. Other features that can be seen in ependymoma include intratumoral hemorrhage, foci of necrosis, cartilaginous and osseous metaplasia, calcification, and hyalinization of blood vessels. Grading: The criteria for distinguishing ependymoma, grade II from anaplastic ependymoma, grade III are not well established and the significance of histologic grading remains controversial [2, 3, 4, 5] . The current WHO criteria states that anaplastic ependymomas, WHO grade III are characterized by increased cellularity, brisk mitotic activity, pseudopalisading necrosis, and microvascular proliferation [1] . However, it is emphasized that necrosis by itself, in the absence of pseudopalisading, does not warrant the diagnosis of anaplasia, since lower grade forms of ependymoma can show degenerative changes. A recent study that investigated univariate histologic features of prognosis for ependymomas identified 4 features associated with aggressive behavior: 1) hypercellularity; 2) vascular proliferation; 3) mitoses > 4/10 HPF; and 4) necrosis [3] . These features were used to construct models of grading that correlated best with clinical outcome. The authors concluded that the presence of two of these four features should be present to make the diagnosis of anaplastic ependymoma, grade III. Other recent studies that have used similar, consistent definitions of anaplasia (requiring hypercellularity, nuclear anaplasia, and vascular proliferation) have also shown a significant survival difference between ependymoma and anaplastic ependymoma [4, 5] . A number of studies have investigated the utility of MIB-1 immunohistochemistry in ependymoma [1, 8] . High MIB-1 proliferation index is associated with anaplasia and poor outcome. However, reproducible cut-off levels that could be used across institutions have not been established; in the literature, they have varied from 4% to 20%. Immunohistochemistry and Electron Microscopy : Since ependymomas are glial neoplasms, generally displaying abundant fibrillarity, it is not surprising that they are GFAP positive in the majority of cases. Nevertheless, the latter can be very useful in highlighting the typical pattern of thin cytoplasmic processes radiating towards blood vessels in the center of perivascular pseudorosettes. They also show reactivity for vimentin and S-100. EMA staining is variable, but when present, is usually seen along the surfaces formed by ependymal cells, either in linear arrays or within the canals of ependymal rosettes. More helpful is a rounded or dot-like pattern of intracytoplasmic staining, corresponding to the intracytoplasmic lumina seen on electron microscopy. More recently, studies have shown that the majority of ependymomas are also CD99 positive, often highlighting not only the cell membrane, but also the same intracytoplasmic lumina seen with EMA [9, 10] . Ultrastructural studies are not needed to establish the diagnosis of ependymoma in cases displaying classic morphologic features. Nevertheless, in unusual or poorly differentiated examples, it is the gold standard, diagnostic features including cilia, microvilli, long "zipper-like" intercellular junctions, and intracellular lumina. Clear Cell Variant of Ependymoma : Clear cell ependymomas (CCE) occur predominantly in children and young adults [11, 12] . This variant is relatively rare and is often misdiagnosed as oligodendroglioma, given the rounded nuclei, "fried egg" clear cell appearance, and even a chicken wire-like vascular pattern in some cases. In contrast to the diffusely infiltrative growth pattern of oligodendrogliomas however, CCE shows the typical solid growth and pushing margins of ependymomas in general; they also do not harbor 1p and 19q deletions (discussed below). Perivascular pseudorosettes can be subtle, although they were well formed in the current case. The immunohistochemical and ultrastructural features of ependymomas in general are also seen in CCE. In comparison to other ependymomas, supratentorial location, cyst formation, and features of anaplasia are particularly common. Although the number of reported cases is still limited, there is some suggestion that pediatric examples are more aggressive than their adult counterparts. Loss of chromosome 18 has recently been reported to be common in the anaplastic CCEs [11] , although it was not seen in the current example. Differential Diagnosis of CNS Tumors with a "Fried Egg" or Clear Cell Appearance Many CNS tumors may have rounded nuclei with clear haloes and often, the first consideration is oligodendroglioma. However, the differential diagnosis is relatively broad (Table below). The greatest overlap is with the diffuse astrocytomas, which generally share a common clinical presentation and infiltrative growth pattern. The current distinction rests primarily on nuclear cytology, which unfortunately is far from perfect [13, 14] . Astrocytic nuclei are more typically oval, spindled, or irregular, although when cut in cross section, they may appear rounded and occasionally associated with clear haloes, particularly in poorly fixed or partially autolyzed specimens, such as those encountered in CUSA (Cavitron ultrasonic aspiration) specimens where the filtered tissue often sits for prolonged periods in saline before being submitted to pathology. GFAP-immunoreactive processes support astrocytic over oligodendroglial derivation, though even this is fairly unreliable, since one may be fooled by processes belonging to entrapped reactive astrocytes or by non-specific background staining of axons. Likewise, there are many oligodendrogliomas with GFAP-positive minigemistocytes and gliofibrillary oligodendrocytes. Thus, much of the remaining diagnostic difficulties reflect the fact that there are currently no absolutely specific oligodendroglioma markers. Though some would argue that 1p/19q testing now fulfills that role (discussed below), the sensitivities and specificities are still imperfect. Other entities are more readily distinguishable based on specific clinical, gross/radiologic, and histopathologic features (Table). Although it is not widely appreciated, pilocytic astrocytoma may display regions that not only resemble diffuse astrocytoma, but oligodendroglioma as well. Fortunately, the clinical features are usually distinctive and a pediatric tumor in the cerebellum, optic pathway, hypothalamus/third ventricle, thalamus, dorsal brainstem, or spinal cord is far more likely to represent pilocytic astrocytoma than oligodendroglioma. Furthermore, the majority of pilocytic astrocytomas harbor at least a few Rosenthal fibers and/or eosinophilic granular bodies (EGBs), though neither is absolutely necessary or specific for the diagnosis. An "intraventricular oligodendroglioma" is a central neurocytoma until proven otherwise and its neuronal differentiation is easily verified with synaptophysin and Neu-N immunostains. Pineocytomas look virtually identical to central neurocytoma, but involve the pineal gland instead. A dysembryoplastic neuroepithelial tumor (DNT) may be impossible to separate from oligodendroglioma on a needle biopsy, but larger specimens typically reveal the characteristic intracortical localization, nodular growth pattern, and "floating neurons". Clear cell ependymoma is discussed above. Genetic Biomarkers in Oligodendrogliomas In no other area of brain tumor pathology has genotyping proven more clinically valuable than in the genetic profiling of oligodendroglial tumors [15] . Comprising 10- 25% of adult gliomas, oligodendrogliomas tend to behave in a less aggressive fashion than astrocytomas, with slower progression and longer patient survival. Likewise, the dramatic therapeutic responses to PCV (procarbazine, CCNU, vincristine) chemotherapy reported in subsets of anaplastic oligodendroglioma patients is a noteworthy finding in comparison to the usual lack of response in astrocytomas. LOH and FISH studies have shown 1p and 19q codeletion in 60-90% of oligodendrogliomas. Cairncross and colleagues' landmark study was the first to establish an association between anaplastic oligodendrogliomas bearing this "molecular signature" and the likelihood of favorable therapeutic response and prolonged survival [16] . Similarly, Smith et al reported that combined 1p/19q deletions were associated with prolonged survival in oligodendrogliomas, including low-grade examples [17] . Studies have further suggested that these "genetically favorable" oligodendrogliomas are also more sensitive to other forms of therapy, including radiation and less toxic chemotherapeutic agents, such as temozolomide [18, 19] . Given the prognostic and therapeutic implications, we and others routinely perform 1p/19q testing in all oligodendrogliomas and tumors with suspected oligodendroglial features. We have reported our initial observations in detail [20] and they've remained valid over time with over 1500 gliomas tested thus far. As in a number of retrospective series, we've found 1p/19q codeletion to be highly associated with morphology: 85% in oligodendrogliomas, 15% in mixed oligoastrocytomas (MOAs), and <1% in astrocytomas (p<0.001). With respect to patient survival, our results have been similar to retrospective studies in that the "genetically favorable" (1p/19q codeletion) pattern is greatly overrepresented in long-term survivors. There have however been notable exceptions and therefore, the genetics should not be interpreted in a vacuum, but rather as a supplement to more classic clinical and pathologic prognosticators. Apart from the obvious prognostic implication discussed above, identification of 1p/19q codeletions may also be helpful in daily pathology, as there are a number of lesions that pose formidable diagnostic challenges. For example, dysembryoplastic neuroepithelial tumors (DNTs), central neurocytomas, extraventricular neurocytomas, and clear cell ependymomas do not have these deletions, with the exception of extraventricular neurocytomas (EVNs). In our series, two of twelve EVNs harbored 1p/19q codeletion, suggesting that either this genetic signature is not entirely specific or that these rare tumors may be histogenetically related to oligodendrogliomas. Reports of oligodendrogliomas with neurocytic differentiation support the latter theory [21] . As discussed in the previous section however, an even more common differential diagnostic consideration is small cell glioblastoma. Compared to their adult counterparts, pediatric oligodendroglial tumors are far less common, with little published data regarding their clinicopathologic or molecular characterization. In Raghavan and colleagues' review of 26 cases, 1p/19q deletions were rare and not obviously associated with clinical outcome [22]. Those with deletions were most commonly teenagers and therefore, may represent older children with "adult type" oligodendroglioma. These observations have been subsequently validated by other groups as well [23]. Table 1 - Differential Diagnosis for Clear Cell Tumors of CNS Diagnosis Helpful Distinguishing Features Clinical Gross/Radiology Histology/Genetics Oligodendrogliomas (WHO II-III) Adult (usually) Superficial/cortical epicenter, extensive calcification Monomorphic round nuclei, crisp nuclear membranes, GFAP+ MGs / gliofibrillary oligos, -1p/19q Diffuse Astrocytomas (WHO II-III) Adult (usually) Deep epicenter Elongated dark nuclei, pleomorphism, GFAP+ processes, TP53 mutations Small Cell GBM (WHO IV) Adult (usually) Deep epicenter, ring-enhancing (~2/3rd) Monomorphic oval nuclei, many mitoses, GFAP+ processes, EGFR-AMP, -10q Pilocytic Astrocytoma (WHO I) Child or young adult Cerebellum, optic pathway, spinal cord, hypothalamus/3rd v., cyst with enhancing mural nodule, discrete Rosenthal fibers, eosinophilic granular bodies (EGBs), limited invasion, thin GFAP+ processes DNT (WHO I) Child or young adult, long seizure history Temporal lobe, limited to cortex, T1-dark, T2-bright nodularities Patterned mucin-rich nodules, floating neurons, oligo-like cells, limited to cortex Central Neurocytoma or Pineocytoma (WHO II) Hydrocephalus-type symptoms Lateral ventricle / septum pellucidum, pineal, enhancing Rosette / neuropil formation, oligo-like cells, synaptophysin+ Clear Cell Ependymoma (WHO II-III) Child or young adult Discrete borders, enhancing, cystic, often supratentorial Non-infiltrative, GFAP+ perivascular pseudorosettes, CD99/EMA+ lumina Clear Cell Meningioma (WHO II) Child or young adult (usually) Spinal cord / post. fossa, extra-axial Dural-based, vague whorling, perivascular / interstitial collagen, glycogen-rich, EMA+ Metastatic Renal Cell Carcinoma Adult Discrete borders, enhancing, single or multiple Non-infiltrative, large nucleoli, CK+, EMA+, CD10+ Hemangioblastoma Adult Cerebellum, spinal, brainstem, nerve root, cyst with enhancing mural nodule Vacuolated inhibin+, S-100+, NSE+ stromal cells, hypervascular, EMH (~10%) Germinoma Child Suprasellar and/or pineal, enhancing Large cells, big nucleoli, lymphocytic infiltrate, granulomas References Kleihues P, Cavenee W, eds. World Health Organization classification of tumours. Pathology and genetics: tumours of the nervous system. (2nd ed). Lyon: IARC Press; 2000. Figarella-Branger D, Civatte M, Bouvier-Labit C, et al. Prognostic factors in intracranial ependymomas in children. J Neurosurg. Oct 2000;93(4):605-613. Ho DM, Hsu CY, Wong TT, Chiang H. A clinicopathologic study of 81 patients with ependymomas and proposal of diagnostic criteria for anaplastic ependymoma. J Neurooncol. Aug 2001;54(1):77-85. Korshunov A, Golanov A, Sycheva R, Timirgaz V. The histologic grade is a main prognostic factor for patients with intracranial ependymomas treated in the microneurosurgical era: an analysis of 258 patients. Cancer. Mar 15 2004;100(6):1230-1237. Merchant TE, Jenkins JJ, Burger PC, et al. Influence of tumor grade on time to progression after irradiation for localized ependymoma in children. Int J Radiat Oncol Biol Phys. May 1 2002;53(1):52-57. Smyth MD, Horn BN, Russo C, Berger MS. Intracranial ependymomas of childhood: current management strategies. Pediatr Neurosurg. Sep 2000;33(3):138-150. Foreman NK, Love S, Thorne R. Intracranial ependymomas: analysis of prognostic factors in a population-based series. Pediatr Neurosurg. 1996;24(3):119-125. Ritter AM, Hess KR, McLendon RE, Langford LA. Ependymomas: MIB-1 proliferation index and survival. J Neurooncol. Oct 1998;40(1):51-57. Choi YL, Chi JG, Suh YL. CD99 immunoreactivity in ependymoma. Appl Immunohistochem Mol Morphol. Jun 2001;9(2):125-129. Kawano N, Yasui Y, Utsuki S, Oka H, Fujii K, Yamashina S. Light microscopic demonstration of the microlumen of ependymoma: a study of the usefulness of antigen retrieval for epithelial membrane antigen (EMA) immunostaining. Brain Tumor Pathol. 2004;21(1):17-21. Fouladi M, Helton K, Dalton J, et al. Clear cell ependymoma: a clinicopathologic and radiographic analysis of 10 patients. Cancer. Nov 15 2003;98(10):2232-2244. Min KW, Scheithauer BW. Clear cell ependymoma: a mimic of oligodendroglioma: clinicopathologic and ultrastructure considerations. Am J Surg Pathol. 1997(21):820–826. Burger PC. What is an oligodendroglioma? Brain Pathol. Apr 2002;12(2):257-259. Perry A. Oligodendroglial neoplasms: current concepts, misconceptions, and folklore. Adv Anat Pathol. Jul 2001;8(4):183-199. Fuller CE, Perry A. Molecular diagnostics in central nervous system tumors. Adv Anat Pathol. Jul 2005;12(4):180-194. Cairncross JG, Ueki K, Zlatescu MC, et al. Specific genetic predictors of chemotherapeutic response and survival in patients with anaplastic oligodendrogliomas. J Natl Cancer Inst. Oct 7 1998;90(19):1473-1479. Smith JS, Perry A, Borell TJ, et al. Alterations of chromosome arms 1p and 19q as predictors of survival in oligodendrogliomas, astrocytomas, and mixed oligoastrocytomas. J Clin Oncol. Feb 2000;18(3):636-645. Bauman GS, Ino Y, Ueki K, et al. Allelic loss of chromosome 1p and radiotherapy plus chemotherapy in patients with oligodendrogliomas. Int J Radiat Oncol Biol Phys. Oct 1 2000;48(3):825-830. Chahlavi A, Kanner A, Peereboom D, Staugaitis SM, Elson P, Barnett G. Impact of chromosome 1p status in response of oligodendroglioma to temozolomide: preliminary results. J Neurooncol. Feb 2003;61(3):267-273. Perry A, Fuller CE, Banerjee R, Brat DJ, Scheithauer BW. Ancillary FISH analysis for 1p and 19q status: preliminary observations in 287 gliomas and oligodendroglioma mimics. Front Biosci. Jan 1 2003;8:a1-9. Perry A, Scheithauer BW, Macaulay RJ, Raffel C, Roth KA, Kros JM. Oligodendrogliomas with neurocytic differentiation. A report of 4 cases with diagnostic and histogenetic implications. J Neuropathol Exp Neurol. Nov 2002;61(11):947-955. Raghavan R, Balani J, Perry A, et al. Pediatric oligodendrogliomas: a study of molecular alterations on 1p and 19q using fluorescence in situ hybridization. J Neuropathol Exp Neurol. May 2003;62(5):530-537. Kreiger PA, Okada Y, Simon S, Rorke LB, Louis DN, Golden JA. Losses of chromosomes 1p and 19q are rare in pediatric oligodendrogliomas. Acta Neuropathol (Berl). Apr 2005;109(4):387-392.