Comparative genomic hybridization detects specific cytogenetic abnormalities in pediatric ependymomas and choroid plexus papillomas

Cancer Genetics and Cytogenetics 136 (2002) 121–125

Comparative genomic hybridization detects specific cytogenetic abnormalities in pediatric ependymomas and choroid plexus papillomas Jacques Grilla,b, Hervé Avet-Loiseauc, Arielle Lellouch-Tubianad, Nicolas Sévenete, Marie-José Terrier-Lacombef, Anne-Marie Vénuatb, François Dozg, Christian Sainte-Roseh, Chantal Kalifaa, Gilles Vassala,b,* a

Department of Pediatric Oncology, Institut Gustave Roussy, 39, rue Camille Desmoulins, 94805 Villejuif Cedex, France Laboratory of Pharmacogenetics and Pharmacotoxicology, CNRS-UMR 8532, Institut Gustave Roussy, Villejuif, France c Laboratory of Hematology, Institute of Biology, Nantes, France d Department of Pathology, Necker Hospital, Paris, France e Laboratory of Genetics and Biology of Pediatric Tumors, INSERM U509, Institut Curie, Paris, France f Department of Pathology, Institut Gustave Roussy, Villejuif, France g Department of Pediatric Oncology, Institut Curie, Paris, France h Department of Pediatric Neurosurgery, Necker Hospital, Paris, France Received 14 August 2001; accepted 3 January 2002

b

Abstract

Pathogenesis and genetic abnormalities of ependymomas are not well known and differential diagnosis with choroid plexus tumors may be difficult when these tumors are located in the ventricles. We analyzed 16 samples of primary pediatric ependymomas and seven choroid plexus tumors for significant gains or losses of genomic DNA, using comparative genomic hybridization (CGH). Four ependymoma samples were obtained after surgery for relapse, including one patient whose tumor was analyzed at diagnosis and at first and second relapses. Three out of 16 ependymomas and none of the choroid plexus tumors appeared normal by CGH. In the remaining ependymomas, the number of regions with genomic imbalance was limited. The most frequent copy number abnormality in ependymomas was 22q loss. In one patient from whom multiple samples could be analyzed during tumor progression, no abnormality was present at diagnosis; gain of chromosome 9 and loss of 6q were observed at first relapse and, at second relapse, additional genomic imbalances were loss of 3p, 10q, and chromosome 15. In choroid plexus tumors, recurrent abnormalities were gains of chromosome 7 and region 12q. The recurrent chromosomal abnormalities were clearly different between ependymomas and choroid plexus papillomas (CPP). Recurrent loss of 22q suggests that this region harbors tumor suppressor genes important in the pathogenesis of ependymomas; however, other pathogenic pathways may exist involving 6q and chromosome 10 losses or gain of 1q and chromosome 9. CPP can be distinguished from ependymoma on the basis of CGH abnormalities. © 2002 Elsevier Science Inc. All rights reserved.

1. Introduction Ependymal neoplasms are rare brain tumors; however, in children they represent the third most frequent brain tumor [1]. In adults, the most frequent location is in the spinal cord, whereas in children these tumors are mostly found in the posterior fossa. There is now increasing evidence that spinal and intracranial ependymomas represent two differ-

* Corresponding author. Tel. 33-1-42-11-49-47; fax 33-1-42-11-52-75. E-mail address: [email protected] (G. Vassal).

ent subsets of tumors [2,3]. The extent of resection is the only prognostic factor regularly reported in the literature [1,4,5]. There are currently no prognostic biological markers available and little is known about the molecular abnormalities that underly the pathogenesis of this tumor. Karyotypic and cytogenetic analyses of ependymomas have lagged far behind studies done in medulloblastoma and astrocytic tumors. For the latter ones, a hypothetical multistep progression of genetic events could be proposed to describe the process of increasing malignancy and at least two distinct progression pattern could be evidenced [6]. The cytogenetic abnormalities in ependymal tumors are scarce

0165-4608/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved. PII: S0165-4608(02)00 5 1 6 - 2

122

J. Grill et al. / Cancer Genetics and Cytogenetics 136 (2002) 121–125

compared to other pediatric brain tumors and the karyotype is often normal [4]. The most frequent cytogenetic abnormality reported in ependymomas is loss of the long arm of chromosome 22, this abnormality being more frequent in adults than in children [7,8]. Ebert et al. found that NF2 gene mutation and 22q loss occurred preferentially in spinal cord ependymomas [2]; however, almost one third of the intracranial ependymomas diagnosed in children show loss of genetic material on this locus also [8]. Other recurrent chromosomal abnormalities have been reported, essentially 6q loss [3,7,9,10]. Due to the lack of sufficient data of most of the studies, no correlation could be made between a given cytogenetic abnormality and prognosis. Choroid plexus tumors are rare neoplasms mainly found in young children under 2 years of age [11,12]. Differential diagnosis between ependymoma and choroid plexus tumors, arising in the same area of the brain, is often difficult, especially when tumor immunohistochemical markers are lacking [13]. Because therapeutic options are quite different in these two tumor types, it is desirable to provide additional diagnostic tools. As regard to the cytogenetics of choroid plexus tumors, most of the studies describe only single cases, except for the report of Bhattarcharjee et al. [14]. In this single study, the most frequent abnormalities found in atypical papilloma were gain of chromosomes 7, 12, and 20. Choroid plexus carcinomas showed hypodiploid karyotypes. The aim of our study was to use the new technology of comparative genomic hybridization (CGH) to define and compare copy number abnormalities in these two tumor types. 2. Materials and methods 2.1. Patient material Primary brain tissue samples were obtained from 13 patients with posterior fossa ependymomas at diagnosis (N  9), at relapse (N  2), or both (N  2) (Table 1). For one patient, the tumor material analyzed was a xenograft growing in athymic mice at diagnosis and the primitive tumor at relapse (samples 22 and 69, respectively). In addition, seven supratentorial choroid plexus tumors were analyzed at diagnosis; there were six choroid plexus papillomas (CPP) and one unclassified tumor (Table 2). The cytogenetists (HAV and AMV) were “blinded” for the pathological diagnosis. None of the patients had metastatic disease. Tissue samples were immediately frozen in liquid nitrogen upon tumor removal and stored at 80C until analysis. All tumor samples were analyzed by a single pathologist (ALT) using the revised World Health Organization classification of 1993. Clinical data of the 13 children (7 boys and 6 girls) with ependymoma are described in Table 1. The median age was 3 years (8 months to 10 years). Among the seven patients with choroid plexus tumors, four were boys and three were girls (Table 2). Their median age at diagnosis was 2 years (2 months to 5 years).

Table 1 Patients’ clinical data (ependymomas) Case no.

Age at diagnosis

Surgery

Adjuvant treatment

PFS

OS

6 22 23 62 88 140 166 195 217 232 258 266 282

6y 3y 6y 2y 5y 21 mo 6y 10 y 3y 38 mo 8 mo 8 mo 10 y

Complete Complete Partial Complete Partial Partial Partial Partial Complete Complete Complete Complete Complete

PFRT PFRT PFRT BB-SFOP BB-SFOP NA PFRT PFRT BB-SFOP NA BB-SFOP BB-SFOP BB-SFOP

9 y 12 mo 12 mo 13 mo 4 mo 48 mo 12 mo 12 mo 19 mo 51 mo 60 mo 8 mo 36 mo

9 y 72 mo 48 mo 50 mo 45 mo 54 mo 35 mo 67 mo 30 mo 51 mo 108 mo 15 mo 36mo

Abbreviations: mo, months; OS, overall survival; PFS, patient-free survival.

2.2. Histology For histological analysis, tissue specimens were fixed in acetic-formalin-ethanol (AFA; Carlo-Erba, Milan, Italy) and embedded in paraffin. Paraffin-embedded sections were then routinely stained with hematoxylin-eosin-saphranin (HES). Appropriate immunohistochemical staining was performed when needed for a differential diagnosis. All the samples were obtained during surgical resection. Histology was done on one half of the tumor samples, the rest were frozen for CGH analysis. Only fragments that were completely tumoral were used for analysis. 2.3. CGH CGH was performed as previously described [15]. Briefly, tumor DNA was extracted using phenol-chloroform technique. Tumor DNA was labeled by nick translation with Texas Red (5TR)-dUTP (Dupont-NEN), whereas normal DNA from healthy male and female donors was labeled with fluorescein isocyanate (FITC)-dUTP. A total of 150 ng of each donor’s DNA and 20 ug of unlabeled Cot-1 DNA (GIBCO BRL) were ethanol precipitated, resuspended in hybrisol VII (Oncor, Gaithersburg, MD, USA), and denatured for 10 minutes. After a 48-hour hybridization at 37C,

Table 2 CGH results of the seven patients with choroid plexus tumors CGH results Case no.

Histology

Gains

Losses

56 191 215 250 252 264 278

Papilloma Papilloma Papilloma Papilloma Papilloma Papilloma CPC/ATTR?

13 — 7, 8, 12, 19 9qter, 15 5, 6, 7, 9q, 12, X 2, 7, 8q, 11, 12, 18 —

— 13 17p — 15 10, 17 22

Abbreviations: —, none.

J. Grill et al. / Cancer Genetics and Cytogenetics 136 (2002) 121–125

slides were washed for 5 minutes in 2 SSC at 73C and examined using an epifluoresence microscope (DMRB, Leica). CGH analysis was performed using Powergene software (PSI, Houston, TX, USA). The over- and under-represented DNA segments were determined by calculating TR: FITC average ratio profiles. Average ratio images were calculated from at least six metaphases and at fixed thresholds. Chromosomal gains were considered when fluorescence ratio exceeded 1.15 and losses were considered for ratios lower than 0.85. Genomic amplification was defined by a red/green ratio 1.5. Telomeric and heterochromatic regions were excluded from analysis, because of frequent hybridization artifacts. Other chromosomal regions also frequently present hybridization artifacts, such as 1p, 16q, 19, 22q, and Y. One possibility was to exclude these regions from analysis; however, some of them (especially chromosome 22) have been shown to be frequently rearranged in ependymomas. We have adopted the following strategy. In cases displaying abnormalities in all these regions, they were excluded from analysis. In cases displaying significant variations (i.e., ratio 0.85 or 1.15), a second hybridization experiment with reverse labeling was performed. If the second hybridization confirmed the first results, they were considered significant and included in the results table. In contrast, if the reverse labeling experiment did not show the same pattern, these regions were excluded from analysis. 2.4. Mutations of hSNF5/INI1 Deletions and mutations of the hSNF5/INI1 gene at the DNA and mRNA level were analyzed by polymerase chain reaction and dHPLC as described previously [16].

3. Results The pathological review confirmed the initial diagnosis of either ependymoma or CPP in all patients but one. This latter patient (sample 278) had a complex tumor showing at least two components: a differentiated papillary component that was positive for KL-1 (cytokeratin), S-100 (calciumbinding protein), EMA (epithelial membrane antigen, cytokeratin) and negative for GFAP (glial fibrillary acidic protein), and another undifferentiated component that was negative for all the markers studied. Given the pattern of immunoreactivity, the two possible diagnoses for this tumor were atypical teratoid-rhabdoid tumor (ATTR) or choroid plexus carcinoma (CPC). CGH was successfully performed for all samples. Table 3 summarizes the copy number changes detected in the 13 patients (and 16 different tumor samples) with ependymoma. Twelve of 15 (80%) samples contained aberrations in at least one chromosomal region. The number of imbalances ranged from 0 to 6 (median  1). The most frequent abnormalities were loss of chromosomes 22 or 6, 23% and 15%, respectively. These two copy number abnormalities

123

were never associated. While chromosome 22 loss was usually isolated, 6q loss was associated with numerous other abnormalities. In addition, gain of 1q, the most frequent chromosome gain, was found in two of 13 patients. No amplification was found at classical amplified loci (MYCC, MYCN, and EGFR). For two patients with ependymoma, we could analyze the tumor at diagnosis and at subsequent relapse(s), (see Table 1 for patient’s data). There was an accumulation of chromosomal imbalances during progression of these tumors. Tumor of patient 22 had a normal karyotype at diagnosis (case 22) and subsequently showed isolated loss of chromosome 22 at relapse (case 69). Tumor of patient 62 had a normal karyotype at diagnosis (case 62) but exhibited gain of chromosome 9 and loss of 6q at first relapse (case 225). At second relapse, the tumor had loss of 3q, 10q, and chromosome 15, in addition to the chromosomal aberrations already present at first relapse. Although copy number abnormalities seem to accumulate during progression (see patients with CGH analysis at diagnosis and at relapse), there was no correlation between the number of chromosome imbalances or their nature and the progression-free survival (PFS) or overall survival (OS) of this unselected group of children with intracranial ependymomas. Mean PFS was 3 years 3 months (range 3 months to 10 years) if 1 or less chromosome imbalance was present and 2 years (range 8 months to 5.5 years) if more than one chromosome imbalance was present at diagnosis (P  NS). Table 2 summarizes the copy number changes detected in the 6 patients with choroid plexus tumors. All six tumors tested contained aberrations in at least one chromosomal region and most frequently gains. The most frequent imbalances were gain of chromosomes 7 and 12 in one half of the tumor samples. These two copy number abnormalities were always associated. No amplification was found at classical amplified loci (MYCC, MYCN, and EGFR). Tumor of patient 278 showed an isolated loss of chromosome 22. In order to establish a precise diagnosis, the hSNF5/INI1 gene was analyzed for mutation or homozygous deletions. No somatic mutation or deletion was found in the tumor. 4. Discussion CGH is increasingly used for the identification of chromosomal imbalances in a wide variety of brain tumors. This technique circumvents some of the limitations associated with conventional cytogenetics, such as the difficulty to obtain metaphases in slowly growing tumor cells. Moreover, CGH does not require fresh tumor material, and small pieces of tumor can be analyzed successfully. Thus, CGH is a powerful technique for cytogenetic profiling of tumor cells for diagnostic and prognostic purposes. In this study, the cytogenetic abnormalities detected by CGH in ependymoma were scarce compared to medulloblastoma/PNET [15,17–20]. For example, in our previous CGH study on medulloblastoma/PNET, we found a median

124

J. Grill et al. / Cancer Genetics and Cytogenetics 136 (2002) 121–125

Table 3 CGH results of the 13 patients with ependymoma CGH results Case no.

Sex

Gains

Losses

6 22 6922bis 23 62 22562bis 27162ter 88 140 166 195 217 232 258 266 282

M F F F M M M M F F M M F M M F

— — — 14, 17q — 9 9 — — 1q, 17 12, 20p — — 1q, 2p22p24, 5, Xp 13q14qter 5, 9, 12, X

17p — 22 — — 6q 3q, 6q, 10q, 15 22 11p — 3, 6, 10, 16 — 22 — — 16, 17

Abbreviations: F, female; M, male; —, none.

of six copy number abnormalities (range 0–13) [15], compared to only one reported here for ependymoma. Recently, two studies reported also a low number of chromosomal imbalances in childhood ependymomas [3,10]. A compilation of the CGH data on childhood intracranial ependymomas from these studies and our work revealed that the most frequent losses were of chromosome 22 and 6q, observed in 7 of 49 and 10 of 49 cases, respectively. Except in one case from Reardon et al.’s study [10], these two chromosome imbalances were never associated (expected prevalence 1/35 or 1.4/49, P  NS). This may, however, suggest that there can be at least two different progression pathways for ependymoma, one involving chromosome 22 loss and one involving 6q loss. In our study, these two imbalances were present at relapse in the two patients that could be analyzed at diagnosis and at relapse. Chromosome 22 loss is more frequent in adults (56%) than in children (28%), according to the available literature [8]. The most frequent gain was on 1q, reported in 10 of 49 cases studied by CGH in the two published reports and in our study [3,10]. This chromosome imbalance was found associated with 6q loss or chromosome 22 loss. The search for tumor suppressor gene(s) on chromosome 22 has failed to show any involvement of the NF2, type 2 gene in intracranial ependymoma; however, mutations have been recently found in spinal ependymoma [2]. In a series of 22 childhood ependymomas, Sévenet et al. [16] did not find any mutations or homozygous deletions of another tumor suppressor gene found on 22q, the hSNF5/INI1 gene involved in chromatin organization. There may be other tumor suppressor genes, however, located on 22q, involved in the progression of intracranial ependymoma. Recurrent losses of 6q have been reported by two other studies with similar frequencies around 20–25% [3,7,10]. Other central nervous system (CNS) tumors have shown this abnormality, for example, gliomas [21,22] and PNET

[9]; however, despite numerous reports of 6q loss in a variety of tumors, also outside of the CNS, no candidate tumor suppressor gene has been identified so far. In one patient (case 271) whose tumor could be analyzed by CGH at diagnosis and at two subsequent relapses, a clear progression pattern could be evidenced. In the progression of this tumor 6q loss and gain of chromosome 9 occurred at first relapse, while at diagnosis, the CGH profile of the tumor was normal. This suggests that, in this case, 6q loss is not an initial event but more likely a promoting event occurring early in the progression of ependymoma. At second relapse, additional chromosomal abnormalities involved loss of 3q, 10q, and chromosome 15. As in other glial neoplasms, where 10q loss is observed only in high-grade tumors, this chromosomal abnormality could be regarded as an end-stage hallmark for aggressive tumors. Alternatively, one could consider the possibility that tumor heterogeneity could have provided similar results. Confirmation studies on the respective role of these chromosome imbalances are indeed justified in these settings, especially because, in astrocytic tumors, 6q loss has been associated to high tumor grade [23]. The cytogenetic profile of choroid plexus tumors clearly differed from the one of ependymomas, although the size of the study could not allow statistical confirmation of this finding. In choroid plexus tumors, gains were more frequent than losses. The two most frequent abnormalities were gain of chromosomes 7 and 12 and these two imbalances were always associated. There have been no other publications on CGH for choroid plexus tumors and karyotypes have been seldom reported. One study however found similarly gain of chromosomes 7 and 12 in children with atypical CPP [14]. Taken together, this report and our study suggest that genes on chromosomes 7 and 12 are involved in the oncogenesis of a large subset of choroid plexus tumors. For one patient with a complex supratentorial tumor, despite the use of extensive histological, immunohistochemical, cytogenetic, and molecular techniques, no definitive diagnosis was obtained. The immunostaining was compatible with an ATTR or a CPC. This tumor was diagnosed in a newborn and progressed rapidly, causing death within 6 months. This aggressive behavior despite subtotal surgery is indicative of an ATTR [24]; however, typical rhabdoid features were not detected on the tumor sample analyzed. Comparative genomic hybridization revealed an isolated loss of chromosome 22, but no mutation or homozygous deletion of hSNF5/INI1 was detected. In the analysis of somatic mutations or homozygous deletions of this gene in several tumor types [16], 4 of 7 ATTR and 4 of 6 CPC had mutations while only 1 out of 6 CPP did so. Although no definitive conclusion on histology could be made for this particular patient, the absence of hSNF5/INI1 mutation is unexpected. One can hypothesize that another way to inactivate the function of the protein produced by the remaining allele may have contributed to the oncogenesis of this tumor. CGH is an efficient way to obtain cytogenetic profiles of tumors with only a small amount of archived material. In

J. Grill et al. / Cancer Genetics and Cytogenetics 136 (2002) 121–125

some cases, it could be used as an adjunct to the available histological technique for the diagnosis of equivocal tumors. While simultaneous gain of chromosomes 7 and 12 is indicative of choroid plexus tumors, 6q loss or 1q gain are in favor of an ependymoma. From this study, and the few other reports in the literature, it could be postulated that there are at least two distinct pathways of oncogenesis of ependymoma, one involving a loss of chromosome 22 and another involving loss of 6q. Further studies are warranted to define if the tumors of one or the other pathway have different clinical behavior. We advocate that larger CGH studies should be undertaken for ependymoma where biological prognostic factors are clearly lacking. Acknowledgments Supported in part by the Association de Recherche sur le Cancer (GV), the Fondation de France, Fédération des Centres de Lutte Contre le Cancer (JG) and the Comité Départemental de Loire-Atlantique de la Ligue Contre le Cancer (HAL). References [1] Hamilton RL, Pollack I. The molecular biology of ependymoma. Brain Pathol 1997;7:807–22. [2] Ebert C, von Haken M, Meyer-Puttlitz B, Wiestler OD, Reifenberger G, Pietsch T, von Deimling A. Molecular genetic analysis of ependymal tumors. NF2 mutations and chromosome 22q loss occur preferentially in intramedullary spinal ependymomas. Am J Pathol 1999; 155:627–32. [3] Hirose Y, Aldape K, Bollen A, James CD, Brat D, Lamborn K, Berger M, Feuerstein BG. Chromosomal abnormalities subdivide ependymal tumors into clinically relevant groups. Am J Pathol 2001; 158:1137–43. [4] Applegate GL, Marymont MH. Intracranial ependymomas: a review. Cancer Invest 1998;16:588–93. [5] Grill J, Le Deley MC, Gambarelli D, Raquin MA, Couanet D, PierreKahn A, Habrand JL, Doz F, Frappaz D, Gentet JC, Edan C, Chastagner P, Kalifa C. Post-operative chemotherapy without irradiation for ependymoma in children under five years of age. A multicenter trial of the French Society of Pediatric Oncology (SFOP). J Clin Oncol 2001;19:1288–96. [6] Watanabe K, Tachibana O, Sata K, Yonekawa Y, Kleihues P, Ohgaki K. Overexpression of the EGF receptor and p53 mutations are mutually exclusive in the evolution of primary and secondary glioblastomas. Brain Pathol 1996;6:217–23. [7] Kramer DL, Parmiter AH, Rorke LB, Sutton LN, Biegel JA. Molecular cytogenetic studies of pediatric ependymomas. J Neurooncol 1998;37:25–33. [8] Mazewski C, Soukup S, Ballard E, Gotwals B, Lampkin B. Karyotype studies in 18 ependymomas with literature review of 107 cases. Cancer Genet Cytogenet 1999;113:1–8.

125

[9] Neumann E, Kalousek DK, Norman MG, Steinbok P, Cochrane DD, Goddard K. Cytogenetic analysis of 109 pediatric central nervous system tumors. Cancer Genet Cytogenet 1993;71:40–9. [10] Reardon DA, Entrekin RE, Sublett J, Ragsdale S, Li H, Boyett J, Kepner JL, Look AT. Chromosome arm 6q loss is the most common recurrent autosomal alteration detected in primary pediatric ependymoma. Genes Chromosomes Cancer 1999;24:230–7. [11] Pencalet P, Sainte-Rose C, Lellouch-Tubiana A, Kalifa C, Brunelle F, Sgouros S, Meyer P, Cinalli G, Zerah M, Pierre-Kahn A, Renier D. Papillomas and carcinomas of the choroid plexus in children. J Neurosurg 1998;88:521–8. [12] Chow E, Reardon DA, Shah AB, Jenkins JJ, Langston J, Heideman RL, Sanford RA, Kun LE, Merchant TE. Pediatric choroid plexus neoplasms. Int J Radiat Oncol Biol Phys 1999;44:249–54. [13] Figarella-Branger D, Lepidi H, Poncet C, Gambarelli D, Bianco N, Rougon G, Pellissier JF. Differential expression of cell adhesion molecules (CAM), neural CAM and epithelial cadherin in ependymomas and choroid plexus tumors. Acta Neuropathol 1995;89:248–57. [14] Bhattacharjee MB, Armstrong DD, Vogel H, Cooley LD. Cytogenetic analysis of 120 primary pediatric brain tumors and literature review. Cancer Genet Cytogenet 1997;97:39–53. [15] Avet-Loiseau H, Venuat AM, Terrier-Lacombe MJ, Lellouch-Tubiana A, Zerah M, Vassal G. Comparative genomic hybridization detects many recurrent imbalances in central nervous system primitive neuroectodermal tumours in children. Br J Cancer 1999;79:1843–7. [16] Sévenet N, Lellouch-Tubiana A, Schofield D, Hoang-Xuan K, Gessler M, Birnbaum D, Jeanpierre C, Jouvet A, Delattre O. Spectrum of hSNF5/INI1 somatic mutations in human cancer and genotype-phenotype correlations. Hum Mol Genet 1999;8:2359–68. [17] Nicholson JC, Ross FM, Kohler JA, Ellison DW. Comparative genomic hybridization and histological variation in primitive neuroectodermal tumours. Br J Cancer 1999;80:1322–31. [18] Nishizaki T, Harada K, Kubota H, Ozaki S, Ito H, Sasaki K. Genetic alterations in pediatric medulloblastomas detected by comparative genomic hybridization. Pediatr Neurosurg 1999;31:27–32. [19] Reardon DA, Michalkiewicz E, Boyett JM, Sublett JE, Entrekin RE, Ragsdale ST, Valentine MB, Behm FG, Li H, Heideman RL, Kun LE, Shapiro DN, Look AT. Extensive genomic abnormalities in childhood medulloblastoma by comparative genomic hybridization. Cancer Res 1997;57:4042–7. [20] Russo C, Pellarin M, Tingby O, Bollen AW, Lamborn KR, Mohapatra G, Collins VP, Feuerstein BG. Comparative genomic hybridization in patients with supratentorial and infratentorial primitive neuroectodermal tumors. Cancer 1999;86:331–9. [21] Bigner SH, McLendon RE, Fuchs H, McKeever PE, Friedman HS. Chromosomal characteristics of childhood brain tumors. Cancer Genet Cytogenet 1997;97:125–34. [22] Jenkins RB, Kimmel DW, Moertel CA, Schultz CG, Scheithauer BW, Kelly PJ, Dewald GW. A cytogenetic study of 53 human gliomas. Cancer Genet Cytogenet 1989;39:253–79. [23] Myiakawa A, Ichimura K, Schmidt EE, Varmeh-Ziaie S, Collins VP. Multiple deleted regions on the long arm of chromosome 6 in astrocytic tumours. Br J Cancer 2000;82:543–9. [24] Burger PC, Yu IT, Tihan T, Friedman HS, Strother DR, Kepner JL, Duffner PK, Kun LE, Perlman EJ. Atypical teratoid/rhabdoid tumor of the central nervous system: a highly malignant tumor of infancy and childhood frequently mistaken for medulloblastoma: a Pediatric Oncology Group study. Am J Surg Pathol 1998;22:1083–92.