Differential Diagnosis of Central Nervous System Tumors: A Critical Examination of Some Recent Immunohistochemical Applications, The

Archives of Pathology & Laboratory Medicine,  Mar 2008  by Edgar, Mark A,  Rosenblum, Marc K

• E-mail
• Print
• Link

Context.-As we write, novel antibodies that may well alter the routine practice of surgical neuropathology are in development, characterization, and the early stages of clinical use. These will be used for purposes of tumor subclas-sification, as prognostic markers, as identifiers of potential therapeutic targets, and as predictors of treatment response.

Objective.-To provide for nonspecialists a critical assessment of the peer-reviewed literature (necessarily colored by our own experience) as it pertains to several im-munohistochemical reagents that have been recently forwarded as adjuncts to the histologic typing of central nervous system tumors.

Data Sources.-We address in these pages only antibodies that are commercially available, that have been the subjects of multiple published series, and that we have had occasion to use in the course of everyday problem solving.

Conclusions.-Discussion concentrates on the use of 4 antibodies: BAF47 in the diagnosis of atypical teratoid/rhabdoid tumor, OCT4 in intracranial germinoma, β-catenin in craniopharyngioma, and NeuN as a marker of neuronal differentiation in neuroepithelial neoplasms.

(Arch Pathol Lab Med. 2008;132:500-509)

EMBRYONAL NEOPLASMS: BAF47 AND THE IDENTIFICATION OF ATYPICAL TERATOID/RHABDOID TUMORS

The atypical teratoid/rhabdoid tumor (AT/RT) shares with rhabdoid tumors of the kidney and soft tissues a predilection for infants and young children, an aggressive clinical biology, and characteristic genetic abnormalities.1,2 The adjectival "rhabdoid," of course, recognizes the presence within these neoplasms of cells having eccentrically positioned, large, and vesicular nuclei, prominent nucleoli, and densely eosinophilic cytoplasm often compacted in a globose, inclusion-like fashion (Figure 1). "Teratoid" components that can be identified in central nervous system (CNS) variants include poorly differentiated neuroepithe-lial elements of small cell type (these may be indistinguishable from medulloblastoma, pineoblastoma, or su-pratentorial primitive neuroectodermal tumor [sPNET]), spindled mesenchymal constituents, squamous nests, and glandular or tubulopapillary formations.1,3 In fact, only a minority (approximately 25%) of AT/RTs are composed solely of rhabdoid cells and these are obscured in a significant subset of cases by vastly predominant primitive neuroectodermal tumor (PNET)-like populations.1,3 There is now compelling genetic evidence (see later) that select neoplasms wholly devoid of rhabdoid forms and quali fying on histologic grounds as medulloblastomas or other embryonal tumors of central neuroepithelial type actually represent "lopsided" AT/RT variants (Figure 2). This differential diagnosis is hardly an academic matter, as AT/RTs are predictably unresponsive to standard regimens used effectively against medulloblastomas and the like but may be controlled in some cases by intensified adjuvant treatment strategies.4,5 That AT/RTs can occur in a setting of heritable genetic predisposition further mandates their identification by surgical pathologists.2 The difficulties practitioners face in doing so are compounded by the rhabdoid cytologic features on display in some medullo-blastomas and PNETs of large cell/anaplastic type, cho-roid plexus carcinomas, high-grade gliomas, meningio-mas, melanomas, metastatic carcinomas, and sarcomas.6

A confident diagnosis of AT/RT is facilitated in the prototypical case by immunolabeling of rhabdoid elements for vimentin, epithelial membrane antigen, smooth muscle actin, and glial fibrillary acidic protein.1,3 Such cells may also exhibit reactivity for cytokeratins, neurofilament proteins, and synaptophysin. Not all cases are so obliging in the matter of antigen display, however, and, as mentioned, rhabdoid cells may not be much in evidence. A major breakthrough in segregating AT/RTs from potential mimickers came with the linkage of these neoplasms (as well as renal and other extrarenal rhabdoid tumors of the pediatric cohort) to inactivating abnormalities of hSNF5/INI1/SMARCB1/BAF47, a tumor suppressor gene on chromosome 22q11.2.2 Commonly designated by the abbreviated hSNF5/INI1 or INI1 alone, this encodes a ubiquitously expressed protein active in chromatin remodeling. INI1 inactivation in the rhabdoid tumor group conforms in pattern to the 2-hit mechanism operative in the disabling of other tumor suppressing genes, loss of 1 copy resulting from partial deletion or monosomy 22 and the second copy suffering a nonsense or frameshift mutation that produces a novel stop codon. Deletions or mutations involving both INI1 copies may also be encountered. Transmitted germline mutations of this gene underlie a familial rhabdoid tumor syndrome characterized by potentially multifocal, neural and extraneural primaries presenting in the first year of life.2

Genetic assessment may certainly facilitate the distinction of AT/RTs from other CNS neoplasms, as the investigations communicated to date indicate that INI1 mutations are largely, if not entirely, restricted to the former.2,7,8 INI1 mutant tumors morphologically acceptable as me-dulloblastomas or sPNETs have been reported,9-11 but such cases have generally exhibited the clinicobiologic features of AT/RTs and it has been reasonably proposed that these be classified and approached as such.12 Choroid plexus tumors (mainly carcinomas) have also been described as occasionally exhibiting INI1 mutation,8,9,10,13 but, as presently discussed, there are grounds for questioning the pathologic diagnosis of the lesions in question. Undeniable, however, is that chromosome 22q deletions/mono-somy have been identified in a variety of neoplasms, including choroid plexus carcinomas,14 that enter in this differential. Chromosome 22q11.2 deletions, in addition, are not detectable in AT/RT cases with uniparental disomy.15 Most problematic in regard to genetic diagnosis is the fact that only 70% to 75% of neoplasms exhibiting the morphologic and immunophenotypic attributes typical of the AT/RT demonstrate INI1 deletion/mutation on karyotypic and molecular genomic appraisal.2 Pertinent to the current discussion is the observation that a significant subset of ostensibly INI1 intact cases nonetheless manifests decreased gene expression on reverse transcriptase-poly-merase chain reaction analysis or undetectable INI1 protein product by Western blot.2 Accordingly, an antibody to the INI1 protein would constitute a potentially more sensitive adjunct to the diagnosis of AT/RT than current genetic assays. Such an antibody is now commercially available in monoclonal form for use on formalin-fixed, paraffinized material and is variously designated as BAF47/SNF5, anti-BAF47 or, simply, BAF47.