Experimental and Molecular Pathology 89 (2010) 169–174

Contents lists available at ScienceDirect

Experimental and Molecular Pathology

j ourna l homepage: www.e lsev ie r.com/ locate /yexmp

Distinct molecular signatures in pediatric infratentorial glioblastomas definedby aCGH

S. Sharma a, A. Free b, Y. Mei b, S.C. Peiper a,b,c,d, Z. Wang a,d, J.K. Cowell a,b,c,⁎a Department of Pathology, Medical College of Georgia, Augusta, GA, USAb MCG Cancer Center, Medical College of Georgia, Augusta, GA, USAc Georgia Cancer Coalition, Medical College of Georgia, Augusta, GA, USAd Department of Pathology, Jefferson Medical College, Philadelphia, PA, USA

⁎ Corresponding author. MCG Cancer Center, MedicaGA, USA.

E-mail address: jcowell@mcg.edu (J.K. Cowell).

0014-4800/$ – see front matter © 2010 Elsevier Inc. Adoi:10.1016/j.yexmp.2010.06.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 9 May 2010Available online 16 July 2010

Keywords:GliomaGlioblastomaPediatric gliomaCGHDNA copy numberMolecular signature

Glioblastomas (GBM) are rare in children, but reportedly have more varied outcome which suggestsdifferences in tumor etiology compared to typical GBM of adults. To investigate this we performed highresolution array comparative genomic hybridization (aCGH) analysis on three pediatric infratentorial GBM,ages 3.5, 7 and 14 years. Two of these tumors occurred in the brainstem and one in the spinal cord. Whilehistologically typical, one brainstem tumor showed mainly pleomorphic astrocytic cells, whereas the otherbrainstem and spinal tumors showed a GFAP positive small cell component. Whole chromosomal gains (#1and #2) and loss (#20) were seen only in the pleomorphic brainstem GBM, which also showed a high level ofsegmental genomic copy number changes. Segmental loss involving chromosome 8 was seen in all threetumors (Chr8;133039446-136869494, Chr8;pter-3581577, and Chr8;pter-30480019 respectively), whereasloss involving chromosome 16 was seen in only 2 cases with small cell components (Chr16;31827239-qterand Chr16;pter-29754532). Segmental gain of chromosome 7 was shared only between the 2 brainstemcases (Chr7;17187166-qter and Chr7;69824947-qter). Chromosome 17 showed segmental gain of 17q in thebackdrop of loss of 17p only in case 1. Segmental gain of chromosome 1q was seen only in case 2. The spinalGBM showed a relatively stable karyotype with a unique loss of Chr19;32848902-qter. None of the frequentlosses, gains and amplifications known to occur in adult GBM were identified, suggesting that pediatricinfratentorial glioblastomas show a molecular karyotype that was more characteristic of pediatric embryonaltumors than adult GBM.

l College of Georgia, Augusta,

ll rights reserved.

© 2010 Elsevier Inc. All rights reserved.

Introduction

Brain tumors are the most common solid tumors in children andincreasing in incidence, accounting for about one-quarter of allchildhood cancer deaths. Yet, in contrast to adults, malignant gliomasare much rarer in the pediatric population (5–10% of childhoodintracranial neoplasms) (Brat et al., 2007; Ganigi et al., 2005; Kleihueset al., 2007; Korshunov et al., 2005; Nakamura et al., 2007; Pollack,1994; Pollack et al., 2002a,b; Pollack et al., 2003a; Pollack et al., 2006).The molecular features of pediatric malignant gliomas have not beenwell characterized, due to their rarity and also in part hampered bythe inability to obtain tissue for diagnosis due to their critical locationoften in the compactly juxtaposed gray and white matter of thebrainstem (Donaldson et al., 2006). A few studies have identifieddifferent molecular genetic events in the de novo GBMs in childrendespite a similar histologic appearance with high-grade gliomas inadults. The latter have distinct molecular pathways associated with

primary and secondary GBMs (Brat et al., 2007; Ganigi et al., 2005;Korshunov et al., 2005; Pollack et al., 2002a; Pollack et al., 2003a;Pollack et al., 2006; Suri et al. 2009). Prognostic correlates have alsobeen found to be different inmalignant pediatric gliomas compared tothose in adults (Pollack et al., 2002b; Pollack et al., 2003b). Theseobservations suggest that the therapeutic approaches currently beingexplored for adult malignant gliomas may not be adaptable topediatric gliomas, and a better understanding of the molecular basisof pediatric high-grade gliomas is required to facilitate identificationof suitable translational therapies for improving their otherwisedismal outcome. In this paper, we present the results of comparativegenomic hybridization on 3 cases of pediatric GBM and discuss thepotential clinical significance of these findings.

Materials and methods

Following receipt of Human Assurance Committee approval, thehistopathology of biopsy tissues frombrain tumorswas reviewed. Threecases of pediatric infratentorial glioblastomas with ages 3.5, 7 and14 years were identified at our institution (Table 1). Representativesections that showed optimal histologic preservation were prepared

Table 1Summary of the major clinicopathologic features.

Main features Case 1 Case 2 Case 3

Age (yrs)/sex 7 M 3.5 M 14 FMain symptoms: - Headache and gait disturbances for 1 week. - Weakness left leg and peri-umbilical

numbness for 1 month.MRI Brainstem tumor Brainstem tumor - Spinal cord: C6-T5 intramedullary tumorMain histopath. features - Pleomorphic - Pleomorphic - Pleomorphic

- Small to intermediate cell component - Small to intermediate cell componentIHC — positive GFAP GFAP GFAPIHC — Ki67 70% 75%IHC — P53 80% 15% 75%IHC — unusual features Synaptophysin positivity in a subset of tumor cellsFISH — 1p36 Not deleted (polysomy) Not deletedFISH — 19q13 Not deleted (polysomy) Not deletedFISH — EGFR Not amplified (polysomy for chr. 7) Not amplified (polysomy for chr. 7)FISH — PTEN Deleted Deleted

170 S. Sharma et al. / Experimental and Molecular Pathology 89 (2010) 169–174

from formalin-fixed, paraffin-embedded (FFPE) biopsy tumor tissuesobtained at the time of initial diagnosis. To investigate genetic changesthat had occurred in the tumor sample from these 3 cases, arraycomparative genomic hybridization (aCGH) was performed. Usingformalin-fixed, paraffin-embedded (FFPE) samples, DNA was extractedusing the WaxFree sample isolation kit (TrimGen). This DNA was thenlabeled with biotin and hybridized to the Affymetrix 250K Sty 1Mapping array using standard procedures (Lo et al., 2008). The overallcall rate for SNPs for these tumors was 75% or greater, which permittedclear visualization of chromosome events. Data was processed asdescribed by Lo et al. (2008) and visualized using Partek Genomics Suitesoftware (Lo et al., 2008).

Results

Two tumors occurred in the brainstem and one in cervico-thoracicspinal cord (Table 1).While histologically typical, case 1with brainstemtumor showed mainly pleomorphic astrocytic cells (Figs. 1A–D),

Fig. 1. A. Case 1: Densely cellular pleomorphic astrocytic neoplasm, with focal necrosis withmarkedly hyperchromatic and pleomorphic overlapping astrocytic cells in a fibrillary backgroand in dense fibrillary network (GFAP×200). 1D. Case 1: The tumor shows a high p53 imm

whereas the other brainstem tumor in case 2 (Figs. 2A–D) and spinalcord tumor in case 3 (Figs. 3A–D) showed a significant GFAP positivesmall to intermediate tumor cell population. FISH analysis performed incases 1 and 2 only demonstrated PTEN deletion without accompanyingEGFR amplification.

Major genomic findings are detailed in Table 2, and illustrated inFig. 4.Whole chromosomal gains (#1 and #2) and loss (#20)were seenonly in case 1with the pleomorphic brainstemGBM,which also showedahigh level of segmental genomic copynumber changes. Segmental lossinvolving chromosome 8 was seen in all three tumors(Chr8;133039446-136869494, Chr8;pter-3581577, and Chr8;pter-30480019 respectively), whereas loss involving chromosome 16 wasseen in only cases 2 and 3, the two cases with small cell components(Chr16;31827239-qter and Chr16;pter-29754532). Segmental gain ofchromosome 7 was shared only between cases 1 and 2, the twobrainstem cases (Chr7;17187166-qter and Chr7;69824947-qter). How-ever, no EGFR amplification was detected. Chromosome 17 showedchanges only in case 1, in the form of segmental loss of 17p but also gain

pseudopalisading (H&E×100). 1B. Case 1: Densely cellular infiltrating neoplasm, withund (H&E×200). 1C. Case 1: The tumor shows diffuse GFAP positivity, both cytoplasmicunohistochemical nuclear labeling index of up to about 80% (p53×200).

Fig. 2. A. Case 2: Densely cellular astrocytic neoplasm, with foci of necrosis with pseudopalisading (H&E×40). 2B. Case 2: Densely cellular infiltrating neoplasm, with markedlyhyperchromatic and pleomorphic astrocytic cells in a fibrillary background, with focal moderate microvascular proliferation (H&E×400). 2C. Case 2: The tumor shows extensiveGFAP positivity, both cytoplasmic and in dense fibrillary network (GFAP×200). 2D. Case 2: The tumor shows a relatively low p53 immunohistochemical nuclear labeling index(p53×200).

171S. Sharma et al. / Experimental and Molecular Pathology 89 (2010) 169–174

of a portion of 17q. Segmental gain of chromosome 1qwas seen only incase 2. The spinal GBM showed a relatively stable karyotype with aunique loss of Chr19;32848902-qter. None of the frequent losses, gainsand amplifications known to occur in adult GBM (Lo et al., 2007b) wereidentified in this sample (see Discussion).

Fig. 3. A. Case 3: Densely cellular pleomorphic astrocytic neoplasm, with focal necrosis witneoplasm, withmarkedly hyperchromatic and pleomorphic overlapping astrocytic cells in a r3C. Case 3: The tumor shows diffuse GFAP positivity with dense fibrillary network in portionlabeling index of up to about 70% (P53×200).

Discussion

The primary/de novo glioblastomas (GBMs), which typically affectolder patients, are characterized by amplification of the epidermalgrowth factor receptor gene (EGFR) in about 35–50%, deletion or

h pseudopalisading (H&E×100). 3B. Case 3: Densely cellular infiltrating portion of theather sparse fibrillary background with focal necrosis, including small cells (H&E×200).s (GFAP×400). 3D. Case 3: The tumor shows a high p53 immunohistochemical nuclear

Table 2Summary of the findings of aCGH analysis demonstrating the exact chromosomallocations of genomic copy number changes.

Tumor Segmental loss Segmental gain WCloss

WCgain

Case 1 Chr3;42510515-63927208 Chr2;231587670-qter Chr;20 Chr;1Chr6;64574384-154498139

Chr4;pter-124601275 Chr;2

Chr8;133039446-136869494

Chr7;17187166-qter

Chr9;pter-67686772 Chr9;67686772-qterChr13;20205851-89391673

Chr15;59115255-qter

Chr14;pter-103680426 Chr17;61674852-qterChr15;32732867-59115255

Chr18;19701149-24816145

Chr17;pter-58683987Chr18;9438161-18515035Chr18;32147201-qter

Case 2 Chr8;pter-3581577 Chr1;120987550-qterChr16;qter-31827239 Chr7;69824947-qter

Chr7;109504442-121465706

Case 3 Chr8;pter-30480019Chr16;pter-29754532Chr16;56160245-88668979Chr19;32848902-qter

Fig. 4. Summary of comparative genomic hybridization analyses describing whole chromosostudied.

172 S. Sharma et al. / Experimental and Molecular Pathology 89 (2010) 169–174

mutation of phosphate and tensin homolog tumor suppressor gene(PTEN) inabout15–40%of adultGBM(Ekstrandet al., 1992;Galanis et al.,1998; He et al., 1995; Ohgaki et al., 2004; Reifenberger and Collins, 2004;Schmidt et al., 1999; Watanabe et al., 1996), and loss of heterozygosity(LOH) of chromosome 10 in about 70% (Fujisawa et al., 2000; Kleihues etal., 2007). The immuno-expression of CDK inhibitors p16 and p27, whichdecreases with increasing tumor grades in adult astrocytomas, has beenidentified in about 57–68% and 54% of adult glioblastomas respectively,and are associated with poor survival (Kirla et al., 2000; Leenstra et al.,1998; Ranuncolo et al., 2004; Suri et al., 2009). In contrast, secondaryglioblastomas developing in younger adults, evolve from pre-existinglow-grade gliomas (Galanis et al., 1998; He et al., 1995; Ohgaki andKleihues, 2007; Reifenberger and Collins, 2004; Suri et al., 2009;Watanabe et al., 1996). These tumors show frequent and additive (1)mutations of the TP53 tumor suppressor gene reflected in p53overexpression as identified by immunohistochemistry in over 60%cases (Pollack et al., 1997; Watanabe et al., 2002), (2) amplification/overexpression of PDGFR-a (Kleihues et al., 2007), (3) LOH onchromosomes 10q, 19q, and 22q (Fujisawa et al., 2000), and (4)promoter methylation of RB1, TIMP-3, and HRK (Nakamura et al., 2005).

The few studies reported in pediatric high-grade gliomas haveidentified different molecular genetic events in the de novo GBMs inchildren compared to adults, despite histologic similarity. EGFRamplification (Bredel et al., 1999; Cheng et al., 1999; Di Sapio et al.,2002; Ichimura et al., 2000; James et al., 1988; Jaros et al., 1992), and

mal and segmental gains (red) and losses (blue) of genomic copy numbers in the 3 cases

173S. Sharma et al. / Experimental and Molecular Pathology 89 (2010) 169–174

PTEN deletion/mutation (Cheng et al., 1999; Kraus et al., 2002; Raffelet al., 1999; Rasheed et al., 1997; Sung et al., 2000; Suri et al., 2009) areinfrequent or absent in pediatric glioblastomas. No EGFR amplificationwas found in the 2 cases with PTEN deletion by FISH in our study.Secondary GBMs are rare in the pediatric population, yet pediatric high-grade gliomas show genetic resemblance to adult secondary GBMs interms of TP53mutations (33–38%) and protein accumulation (35–75%)(Cheng et al., 1999; Ichimura et al., 2000; Pollack et al., 2002a; Suri et al.,2009; Watanabe et al., 2002). Although the frequency of p53 proteinexpression (63%) is comparable (Suri et al., 2009), the reportedincidence of p53 mutation (27%) is much less than that (60%–80%) inadult diffuse astrocytomas and secondary glioblastomas (Nakamuraet al., 2007). Moreover p53 mutations in pediatric GBMs are associatedwith adverse prognosis (Pollack et al., 2006). P53 immuno-expressionwas noted in all 3 cases in our study, although it wasmore conspicuousin cases 1 and 3. P53mutation analysis was not performed in our study.

Astrocytic tumors of both adults and children share at least somemolecularmechanismsof tumorprogression. For example, genomic losson 10q, a common chromosomal change in 30% pediatric high-gradeastrocytomas (Nakamura et al., 2007), was less frequent than that inadult primary glioblastomas but is closer to secondary glioblastomas ofadults. Similarly LOH at 22q in 61% of pediatric GBMwas comparable to53% in adult GBM, andwas noted across all histologic grades (Nakamuraet al., 2007). The common deleted region in 22q12.3 spanned theD22S1176 and D22S1172 interval and includes the TIMP-3 tumorsuppressor gene, which is also frequently deleted in secondaryglioblastomas (Nakamura et al., 2007). In the rare reported cases ofEGFR amplification in pediatric GBMs, the tumorwas supratentorial andhemispheric, and occurred at a much lower frequency than in adulttumors (Bredel et al., 1999). As in adult tumors, the reported rare caseswith EGFR amplification did not show p53mutations in pediatric GBMs(Nakamura et al., 2007). Moreover, similar to adult secondaryglioblastomas, PTENmutations detected in only two (11%) glioblastomain one study (Nakamura et al., 2007, did not correlate with EGFRamplification. PDGFRA gene amplification was detected in only 1 of 38gliomaswithout EGFR amplification (Di Sapio et al., 2002) andmay alsohave a role in tumor progression. LOH at 1p/19q was significantly lesscommon in pediatric tumors (Nakamura et al., 2007) compared to adultlesions of similar size and malignancy (Watanabe et al., 2002).

The characteristic range of cytogenetic abnormalities which occur athigh frequency in adult GBM, such as loss of chromosome 10, deletion ofthe p16 gene in 9p21 (oftenhomozygous) amplification of the EGFR locusand gain of chromosome 7 (Lo et al., 2008), was not seen in the pediatricGBM cases described in this study. Similar conclusions were reached inanother study of pediatric GBM (Sanders et al., 2007). Although still asmall sample size, because of the rare nature of these tumors, our studyset provided a relatively unique genomic profile in pediatric GBMs. Thisincluded whole chromosomal gains (#1 and #2) and loss (#20) only inthe pleomorphic brainstem GBM, which also showed a high level ofsegmental genomic copy number changes. Segmental loss involvingchromosome 8 was seen in all three tumors (Chr8;133039446-136869494, Chr8;pter-3581577, and Chr8;pter-30480019 respectively),whereas loss involving chromosome 16 was seen in only 2 cases withsmall cell components (Chr16;31827239-qter and Chr16;pter-29754532). Although segmental losses involving 8p, 16p and 16q arereported occasionally in adult GBM, they are infrequent. Segmental gainof chromosome 7 was shared only between 2 brainstem cases(Chr7;17187166-qter andChr7;69824947-qter). The spinalGBMshoweda relatively stable karyotype with a unique loss of Chr19;32848902-qter.Losses involving the long arm of chromosome 19 are also infrequent inadult GBM and more typically found in anaplastic oligodendroglioma(Ranuncolo et al., 2004). None of the frequent losses, gains andamplifications known to occur in adult GBM (Lo et al., 2007a,b) wereidentified in this sample. In fact, losses involving 8p and 16q, and gains of7 and 17 are far more typical of medulloblastomas (Lo et al., 2007a; Rossiet al., 2006). Moreover, segmental gain of chromosomes 17q (in the

backdrop of loss of 17p) and 1q, as noted in cases 1 and 2 respectively,have recently been associated with adverse outcome in medulloblasto-mas (Pfister et al., 2009). From amolecular genetic standpoint, therefore,thepediatricGBMdescribedhere ismore typical of pediatric brain tumorsthan adult GBM.

Clinical studies of high-grade astrocytomas in young children havefound a more favorable prognosis than similar tumors in older patients,despite frequent recurrences, long-term complications, and delay oravoidance of irradiation, suggesting biologic differences from the moreaggressive tumors encountered in older patients (Dufour et al., 2006;Sanders et al., 2007). It has also been suggested that GBM in youngerchildren may possibly be associated with distinct chromosomeabnormalities compared with GBM in older children (Nakamura et al.,2007). This impression is based on a higher frequency of p53 mutationsandLOHon19q and22q in tumors fromchildren six ormore years of ageat diagnosis compared to younger children (Nakamura et al., 2007;Pollack et al., 1997). Moreover, GBMs from the older children showed22q LOH at distal region 22q12.3–13.31 (Nakamura et al., 2007), similarto adult primary glioblastomas (Nakamura et al., 2005). These findings,albeit on small number of cases, indicate that although it may not bepossible to clearly define genetic pathways in pediatric gliomaprogression, age-related differences might exist. The current study,albeit on a small number of cases, does not provide any data to supportthis hypothesis.

Caveats of the present study include the small sample sizewhich didnot allow us to draw any statistical conclusions. A caveat of thehistopathologic evaluationwas the lack of EGFR immunostaining. It hasbeen noted that despite the infrequency of EGFR amplification inpediatric malignant glioma, immunohistochemical overexpression ofEGFR has been noted in a subset of cases, presumably reflectingupregulation of receptor expression by mechanisms other thanamplification (Pollack et al., 2006). Similarly, the low frequency ofPTEN mutation or deletion can be explained by activation of Akt or itsdownstreammediators by othermolecular events,whichmay influenceresponse to signaling inhibitors (Pollack et al., 2006). Interestingly, it hasrecently been observed that adult malignant gliomas expressing theEGFRvIII mutation, but with intact PTEN (thereby lacking constitutiveAkt activation), are much more likely to show regression followingtreatment with EGFR kinase inhibitors (Mellinghoff et al., 2005). Thisfinding suggests a potential role of evaluating EGFR-targeted therapiesin pediatric malignant gliomas, based on immunohistochemical EGFRexpression, rather than FISH assessment of EGFR amplification. EGFRimmunostaining was not performed in any of our cases, as it has oftenbeen found to provide non-contributory results. The discrepancybetween presence of PTEN deletion in cases 1 and 2 on FISH and itslack of detection on aCGH is likely a function of intratumoralheterogeneity, a common finding in glioblastomas.

To conclude, this FFPE based, high resolution DNA copy numberprofiling in the examined subset of pediatric infratentorial glioblasto-mas showedadistinctmolecular karyotype thatwasmore characteristicof pediatric embryonal tumors than adult GBM.

References

Brat, D.J., Shehata, B.M., Castellano-Sanchez, A.A., et al., 2007. Congenital glioblastoma:a clinicopathologic and genetic analysis. Brain Pathol. 17, 276–281.

Bredel, M., Pollack, I.F., Hamilton, R.L., James, C.D., 1999. Epidermal growth factorreceptor expression and gene amplification in high-grade non-brainstem gliomasof childhood. Clin. Cancer Res. 5, 1786–1792.

Cheng, Y., Ng, H.K., Zhang, S.F., et al., 1999. Genetic alterations in pediatric high-gradeastrocytomas. Hum. Pathol. 30, 1284–1290.

Di Sapio, A., Morra, I., Pradotto, L., Guido, M., Schiffer, D., Mauro, A., 2002. Moleculargenetic changes in a series of neuroepithelial tumors of childhood. J. Neurooncol.59, 117–122.

Donaldson, S.S., Laningham, F., Fisher, P.G., 2006. Advances toward an understanding ofbrainstem gliomas. J. Clin. Oncol. 24, 1266–1272.

Dufour, C., Grill, J., Lellouch-Tubiana, A., et al., 2006. High-grade glioma in childrenunder 5 years of age: a chemotherapy only approachwith the BBSFOP protocol. Eur.J. Cancer 42, 2939–2945.

174 S. Sharma et al. / Experimental and Molecular Pathology 89 (2010) 169–174

Ekstrand, A.J., Sugawa, N., James, C.D., Collins, V.P., 1992. Amplified and rearrangedepidermal growth factor receptor genes in human glioblastomas reveal deletions ofsequences encoding portions of the N- and/or C-terminal tails. Proc. Natl Acad. Sci.USA 89, 4309–4313.

Fujisawa, H., Reis, R.M., Nakamura, M., Colella, S., Yonekawa, Y., Kleihues, P., Ohgaki, H.,2000. Loss of heterozygosity on chromosome 10 is more extensive in primary (denovo) than in secondary glioblastomas. Lab Invest 80, 65–72.

Galanis, E., Buckner, J., Kimmel, D., et al., 1998. Gene amplification as a prognosticfactor in primary and secondary high-grade malignant gliomas. Int. J. Oncol. 13,717–724.

Ganigi, P.M., Santosh, V., Anandh, B., Chandramouli, B.A., Sastry Kolluri, V.R., 2005.Expression of p53, EGFR, pRb and bcl-2 proteins in pediatric glioblastomamultiforme: a study of 54 patients. Pediatr. Neurosurg. 41, 292–299.

He, J., Olson, J.J., James, C.D., 1995. Lack of p16INK4 or retinoblastoma protein (pRb), oramplification-associated overexpression of cdk4 is observed in distinct subsets ofmalignant glial tumors and cell lines. Cancer Res. 55, 4833–4836.

Ichimura, K., Bolin, M.B., Goike, H.M., Schmidt, E.E., Moshref, A., Collins, V.P., 2000.Deregulation of the p14ARF/MDM2/p53 pathway is a prerequisite for humanastrocytic gliomas with G1–S transition control gene abnormalities. Cancer Res. 60,417–424.

James, C.D., Carlbom, E., Dumanski, J.P., et al., 1988. Clonal genomic alterations in gliomamalignancy stages. Cancer Res. 48, 5546–5551.

Jaros, E., Perry, R.H., Adam, L., et al., 1992. Prognostic implications of p53 protein,epidermal growth factor receptor, and Ki-67 labeling in brain tumors. Br. J. Cancer66, 373–385.

Kirla, R.M., Salminen, E.K., Huhtala, S., et al., 2000. Prognostic value of the expression oftumor suppressor genes p53, p21, p16 and pRb, and Ki-67 labelling in high gradeastrocytomas treated with radiotherapy. J. Neurooncol. 46, 71–80.

Kleihues, P., Burger, P.C., Aldape, K.D., Brat, D.J., Biernat, W., Bigner, D.D., 2007.Glioblastoma. In: Louis, D.N., Ohgaki, H., Wiestler, O.D., Cavenee, W.K. (Eds.), WHOClassification of Tumours of the Central Nervous System. IARC Press, Lyon, France,pp. 33–49.

Korshunov, A., Sycheva, R., Gorelyshev, S., Golanov, A., 2005. Clinical utility offluorescence in situ hybridization (FISH) in nonbrainstem glioblastomas ofchildhood. Mod. Pathol. 18, 1258–1263.

Kraus, J.A., Felsberg, J., Tonn, J.C., Reifenberger, G., Pietsch, T., 2002. Molecular geneticanalysis of the TP53, PTEN, CDKN2A, EGFR, CDK4 and MDM2 tumour-associatedgenes in supratentorial primitive neuroectodermal tumours and glioblastomas ofchildhood. Neuropathol. Appl. Neurobiol. 28, 325–333.

Leenstra, S., Oskam, N.T., Bijleveld, E.H., Bosch, D.A., Troost, D., Hulsebos, T.J., 1998.Genetic sub-types of human malignant astrocytoma correlate with survival. Int. J.Cancer 79, 159–165.

Lo, K.C., Bailey, D., Burkhardt, T., et al., 2008. Comprehensive analysis of loss ofheterozygosity events in glioblastoma using the 100 K SNP mapping arrays andcomparison with copy number abnormalities defined by BAC array comparativegenomic hybridization. Genes Chromosom. Cancer 47, 221–237.

Lo, K.C., Rossi, M.R., Eberhart, C.G., et al., 2007a. Genome wide copy numberabnormalities in pediatric medulloblastomas as assessed by array comparativegenome hybridization. Brain Pathol. 17, 282–296.

Lo, K.C., Rossi, M.R., LaDuca, J., et al., 2007b. Candidate glioblastoma development geneidentification using concordance between copy number abnormalities and geneexpression level changes. Genes Chromosom. Cancer 46, 875–894.

Mellinghoff, I.K., Wang, M.Y., Vivanco, I., Haas-Kogan, D.A., Zhu, S., Dia, E.Q., et al., 2005.Molecular determinants of the response of glioblastomas to EGFR kinase inhibitors.N Engl J. Med. 353, 2012–2024.

Nakamura, M., Ishida, E., Shimada, K., Kishi, M., Nakase, H., Sakaki, T., Konishi, N., 2005.Frequent LOH on 22q12.3 and TIMP-3 inactivation occur in the progression tosecondary glioblastomas. Lab. Invest. 85, 165–175.

Nakamura, M., Shimada, K., Ishida, E., et al., 2007. Molecular pathogenesis of pediatricastrocytic tumors. Neuro-Oncology 9, 113–123.

Ohgaki, H., Dessen, P., Jourde, B., et al., 2004. Genetic pathways to glioblastoma: apopulation-based study. Cancer Res. 64, 6892–6899.

Ohgaki, H., Kleihues, P., 2007. Genetic pathways to primary and secondaryglioblastoma. Am. J. Pathol. 170, 1445–1453.

Pfister, S., Remke, M., Benner, A., et al., 2009. Outcome prediction in pediatricmedulloblastoma based on DNA copy-number aberrations of chromosomes 6q and17q and the MYC and MYCN loci. J. Clin. Oncol. 27, 1627–1636.

Pollack, I.F., 1994. Current concepts: brain tumors in children. N Engl J. Med. 331,1500–1507.

Pollack, I.F., Hamilton, R.L., Finkelstein, S.D., Campbell, J.W., Martinez, A.J., Sherwin, R.N.,Bozik, M.E., Gollin, S.M., 1997. The relationship between TP53 mutations andoverexpression of p53 and prognosis in malignant gliomas of childhood. CancerRes. 57, 304–309.

Pollack, I.F., Finkelstein, S.D., Woods, J., et al., 2002a. Expression of p53 and prognosis inchildren with malignant gliomas. N Engl J. Med. 346, 420–427.

Pollack, I.F., Hamilton, R.L., Burnham, J., et al., 2002b. The impact of proliferation indexon outcome in childhood malignant gliomas: results in a multi-institutional cohort.Neurosurgery 50, 1238–1245.

Pollack, I.F., Finkelstein, S.D., Burnham, J., et al., 2003a. The association betweenchromosome 1p and 19q loss and outcome in pediatric malignant gliomas: resultsfrom the CCG-945 cohort. Pediatr. Neurosurg. 39, 114–121.

Pollack, I.F., Boyett, J.M., Yates, A.J., et al., 2003b. The influence of central review onoutcome associations in childhood malignant gliomas: results from the CCG-945experience. Neuro-Oncology 5, 197–207.

Pollack, I.F., Hamilton, R.L., James, C.D., Children's Oncology Group, 2006. Rarity of PTENdeletions and EGFR amplification in malignant gliomas of childhood: results fromthe Children's Cancer Group 945 cohort. J. Neurosurg. 105 (suppl), 418–424.

Raffel, C., Frederick, L., O'Fallon, J.R., et al., 1999. Analysis of oncogene and tumorsuppressor gene alterations in pediatric malignant astrocytomas reveals reducedsurvival for patients with PTEN mutations. Clin. Cancer Res. 5, 4085–4090.

Ranuncolo, S.M., Varela, M., Morandi, A., et al., 2004. Prognostic value of Mdm2, p53 andp16 in patients with astrocytomas. J. Neurooncol. 68, 113–121.

Rasheed, B.K., Stenzel, T.T., McLendon, R.E., et al., 1997. PTEN genemutations are seen inhigh-grade but not low-grade gliomas. Cancer Res. 57, 4187–4190.

Reifenberger, G., Collins, V.P., 2004. Pathology and molecular genetics of astrocyticgliomas. J. Mol. Med. 82, 656–670.

Rossi, M.R., Conroy, J., McQuaid, D., Nowak, N.J., Rutka, J.T., Cowell, J.K., 2006. Array CGHanalysis of pediatric medulloblastomas. Genes Chromosom. Cancer 45, 290–303.

Sanders, R.P., Kocak, M., Burger, P.C., Merchant, T.E., Gajjar, A., Broniscer, A., 2007. High-grade astrocytoma in very young children. Pediatr. Blood Cancer 49, 888–893.

Schmidt, E.E., Ichimura, K., Goike, H.M., Moshref, A., Liu, L., Collins, V.P., 1999.Mutational profile of the PTEN gene in primary human astrocytic tumors andcultivated xenografts. J. Neuropathol. Exp. Neurol. 58, 1170–1183.

Sung, T., Miller, D.C., Hayes, R.L., Alonso, M., Yee, H., Newcomb, E.W., 2000. Preferentialinactivation of the p53 tumor suppressor pathway and lack of EGFR amplificationdistinguish de novo high-grade pediatric astrocytomas from de novo adultastrocytomas. Brain Pathol. 10, 249–259.

Suri, V., Das, P., Jain, A., Sharma, M.C., Borkar, S.A., Suri, A., Gupta, D., Sarkar, C., 2009.Pediatric glioblastomas: a histopathological and molecular genetic study. Neuro-Oncology 11, 274–280.

Watanabe, K., Tachibana, O., Sato, K., Yonekawa, Y., Kleihues, P., Ohgaki, H., 1996.Overexpression of the EGF receptor and p53 mutations are mutually exclusive inthe evolution of primary and secondary glioblastomas. Brain Pathol. 6, 217–224.

Watanabe, T., Nakamura, M., Kros, J.M., Burkhard, C., Yonekawa, Y., Kleihues, P., Ohgaki,H., 2002. Phenotype versus genotype correlation in oligodendrogliomas and low-grade diffuse astrocytomas. Acta Neuropathol. (Berl.) 103, 267–275.