Molecular characterization of disseminated pilocytic astrocytomas.

AIM: Pilocytic astrocytomas represent the most common paediatric tumours of the central nervous system. Dissemination through the ventricular system occurs rarely in patients with pilocytic astrocytomas; however, it is more common in infants with diencephalic tumours, and is associated with a poor outcome. Despite histological similarities with classic pilocytic astrocytomas, it is still unclear whether disseminated pilocytic astrocytomas may have specific molecular features. METHODS: Seventeen disseminated pilocytic astrocytomas were investigated using the molecular inversion probe array and screened for the presence of gene fusions (KIAA1549-BRAF) and mutations (BRAF, RAS and FGFR1). RESULTS: Along with evidence of a constitutive MAPK activation in all cases, the molecular inversion probe array, fluorescence in situ hybridization analysis and mutational study revealed KIAA1549-BRAF fusions in 66% and BRAF(V600E) mutations in 5% of cases. No KRAS, HRAS, NRAS or FGFR1 mutations were found. CONCLUSIONS: disseminated pilocytic astrocytomas showed genetic features similar to classic pilocytic astrocytoma, including a similar incidence of KIAA1549-BRAF fusions, BRAF mutations and a stable genetic profile. Given common activation of the MAPK pathway, the use of specific inhibitors can be hypothesized for the treatment of disseminated pilocytic astrocytomas, along with standard chemo- and/or radiotherapy.

Genomic differences in benign and malignant follicular thyroid tumours using 1-Mb array-comparative genomic hybridisation.

Currently there is a lack of objective markers that can reliably differentiate benign and malignant follicular thyroid tumours. Such markers are needed to avoid the morbidity and cost of diagnosing these lesions by a thyroid lobectomy and then a second operation to remove the remaining half of thyroid if cancer is found. The aim of this research was to look for genomic markers that might solve this important problem. Ethical approval for the project was obtained. DNA was extracted from formalin-fixed paraffin-embedded specimens and copy number analysed using an in-house produced 1-megabase genomic array by comparative genomic hybridization (1Mb-aCGH). Acceptable quality data were obtained in 25/26 (96 %) of adenomas and 17/28 (61 %) of carcinomas. Among the carcinomas, 11 were minimally invasive (MI), 5 widely invasive (WI) and there was one metastasis. Recurrent copy number changes distinguishing benign and malignant included +1p34.2-36.33, +1q, +13q12.11-14.3, +14q22.1-32.33, +20q and -22. +20q became more sensitive (36.4 %) for MI carcinomas, whereas +13q12.11-14.3 and +14q22.1-32.33 became more sensitive (66.7 %) for identifying WI cancers from adenomas. Only in the context of aneuploidy (3 adenomas, 3 MI, 3 WI) there were some specific copy number changes that could differentiate all aneuploid adenomas from carcinomas. This research is the first using 1Mb-aCGH to study benign and malignant follicular thyroid tumours. Overall, the incidence of any copy number changes is low, but there are a number of changes associated with different tumour types. Further research with a larger sample and better quality DNA will clarify these findings.

Evidence for the complete inactivation of the NF2 gene in the majority of sporadic meningiomas.

Meningiomas are common central nervous system tumours which present usually in the 4th and 5th decades of life. Loss of constitutional heterozygosity on chromosome 22 in 60% of sporadic meningiomas has implied the involvement of a tumour suppressor gene. The neurofibromatosis type 2 gene (NF2), a prime candidate for involvement in meningioma, was screened for point mutations. After examining eight of the 16 known NF2 exons in 151 meningiomas, 24 inactivating mutations were characterized. Significantly, these aberrations were exclusively detected in tumours which lost the other chromosome 22 allele. These results provide strong evidence that the suppressor gene on chromosome 22, frequently inactivated in meningioma, is the NF2 gene, and suggest that another gene is involved in the development of 40% of meningiomas.

Fluorescence-guided surgical sampling of glioblastoma identifies phenotypically distinct tumour-initiating cell populations in the tumour mass and margin.

BACKGROUND: Acquiring clinically annotated, spatially stratified tissue samples from human glioblastoma (GBM) is compromised by haemorrhage, brain shift and subjective identification of 'normal' brain. We tested the use of 5-aminolevulinic acid (5-ALA) fluorescence to objective tissue sampling and to derive tumour-initiating cells (TICs) from mass and margin. METHODS: The 5-ALA was administered to 30 GBM patients. Samples were taken from the non-fluorescent necrotic core, fluorescent tumour mass and non-fluorescent margin. We compared the efficiency of isolating TICs from these areas in 5-ALA versus control patients. HRMAS (1)H NMR was used to reveal metabolic alterations due to 5-ALA. We then characterised TICs for self-renewal in vitro and tumorigenicity in vivo. RESULTS: The derivation of TICs was not compromised by 5-ALA and the metabolic profile was similar between tumours from 5-ALA patients and controls. The TICs from the fluorescent mass were self-renewing in vitro and tumour-forming in vivo, whereas TICs from non-fluorescent margin did not self-renew in vitro but did form tumours in vivo. CONCLUSION: Our data show that 5-ALA does not compromise the derivation of TICs. It also reveals that the margin contains TICs, which are phenotypically different from those isolated from the corresponding mass.

CpG island hypermethylation of the neurofibromatosis type 2 (NF2) gene is rare in sporadic vestibular schwannomas.

AIMS: Loss of both wild-type copies of the neurofibromatosis type 2 (NF2) gene is found in both sporadic and neurofibromatosis type 2-associated vestibular schwannomas (VS). Previous studies have identified a subset of VS with no loss or mutation of NF2. We hypothesized that methylation of NF2 resulting in gene silencing may play a role in such tumours. METHODS: Forty sporadic VS were analysed by array comparative genomic hybridization using 1 Mb whole genome and chromosome 22 tile path arrays. The NF2 genes were sequenced and methylation of NF2 examined by pyrosequencing. RESULTS: Monosomy 22 was the only recurrent change found. Twelve tumours had NF2 mutations. Eight tumours had complete loss of wild-type NF2, four had one mutated and one wild-type allele, 11 had only one wild-type allele and 17 showed no abnormalities. Methylation analysis showed low-level methylation in four tumours at a limited number of CpGs. No high-level methylation was found. CONCLUSIONS: This study shows that a significant proportion of sporadic VS (>40%) have unmethylated wild-type NF2 genes. This indicates that other mechanisms, yet to be identified, are operative in the oncogenesis of these VSs.

A20 deletion is associated with copy number gain at the TNFA/B/C locus and occurs preferentially in translocation-negative MALT lymphoma of the ocular adnexa and salivary glands.

The genetic basis of MALT lymphoma is largely unknown. Characteristic chromosomal translocations are frequently associated with gastric and pulmonary cases, but are rare at other sites. We compared the genetic profiles of 33 ocular adnexal and 25 pulmonary MALT lymphomas by 1 Mb array-comparative genomic hybridization (CGH) and revealed recurrent 6q23 losses and 6p21.2-6p22.1 gains exclusive to ocular cases. High-resolution chromosome 6 tile-path array-CGH identified NF-kappaB inhibitor A20 as the target of 6q23.3 deletion and TNFA/B/C locus as a putative target of 6p21.2-22.1 gain. Interphase fluorescence in situ hybridization showed that A20 deletion occurred in MALT lymphoma of the ocular adnexa (8/42=19%), salivary gland (2/24=8%), thyroid (1/9=11%) and liver (1/2), but not in the lung (26), stomach (45) and skin (13). Homozygous deletion was observed in three cases. A20 deletion and TNFA/B/C gain were significantly associated (p<0.001) and exclusively found in cases without characteristic translocation. In ocular cases, A20 deletion was associated with concurrent involvement of different adnexal tissues or extraocular sites at diagnosis (p=0.007), a higher proportion of relapse (67% versus 37%) and a shorter relapse-free survival (p=0.033). A20 deletion and gain at TNFA/B/C locus may thus play an important role in the development of translocation-negative MALT lymphoma.

Real-time quantitative polymerase chain reaction (qPCR) analysis with fluorescence resonance energy transfer (FRET) probes reveals differential expression of the four ERBB4 juxtamembrane region variants between medulloblastoma and pilocytic astrocytoma.

AIMS: We report a comparative study on the mRNA expression of ErbB receptor tyrosine kinases, and in particular ERBB4 transcript variants, in two common paediatric brain tumours: medulloblastoma (MB) and pilocytic astrocytoma (PA). METHODS: While the conventional real-time quantitative polymerase chain reaction was used to measure the expression of ERRBs and ErbB4-processing protease genes, the LightCycler fluorescence resonance energy transfer probes were specifically designed to investigate all of the known ERBB4 juxtamembrane (JM) and cytoplasmic transcript variants. RESULTS: The overall expression of ERBBs suggests that ErbB2/ErbB4 heterodimers and ErbB4 homodimers may be major functional units of the ErbBs in MB, while ErbB2/ErbB3 heterodimers may play a more prominent role in addition to ErbB4-containing dimers in PA. Different expression patterns of ERBB4 JM transcripts in MB, PA and normal brain were observed. The JM-d variant was only detected in MBs, while JM-c was present in MB and PA but was not identified in normal brain. The expression of cleavable ERBB4 transcript variants was elevated in PAs and MBs compared with normal brain, while mRNA levels of ErbB4-processing proteases were similar in both tumour types and normal brain. This suggests that proteolytic cleavage of ErbB4 may be more common in MB and PA, which leads to signalling events divergent from those in normal brain. CONCLUSION: Taken together, these results suggest that ErbB4 processing and function may be altered in brain tumours, such as MB and PA, via differential expression of JM transcript variants.

p53-independent mechanisms regulate the P2-MDM2 promoter in adult astrocytic tumours.

The MDM2 gene is amplified and/or overexpressed in about 10% of glioblastomas and constitutes one of a number of ways the p53 pathway is disrupted in these tumours. MDM2 encodes a nuclear phosphoprotein that regulates several cell proteins by binding and/or ubiquitinating them, with p53 being a well-established partner. MDM2 has two promoters, P1 and P2 that give rise to transcripts with distinct 5' untranslated regions. Transcription from P2 is believed to be controlled by p53 and a single-nucleotide polymorphism (SNP309, T>G) in P2 is reported to be associated with increased risk for, and early development of, malignancies. The use of P1 and P2 has not been investigated in gliomas. We used RT-PCR to study P1- and P2-MDM2 transcript expression in astrocytic tumours, xenografts and cell lines with known MDM2, TP53 and p14(ARF) gene status. Both promoters were used in all genetic backgrounds including the use of the P2 promoter in TP53 null cells, indicating a p53-independent induction of transcription. Transcripts from the P1 promoter formed a greater proportion of the total MDM2 transcripts in tumours with MDM2 amplification, despite these tumours having two wild-type TP53 alleles. Examination of SNP309 in glioblastoma patients showed a borderline association with survival but no apparent correlation with age at diagnosis nor with TP53 and p14(ARF) status of their tumours. Our findings also indicate that elevated MDM2 mRNA levels in tumours with MDM2 amplification are preferentially driven by the P1 promoter and that the P2 promoter is not only regulated by p53 but also by other transcription factor(s).

A PDGFRA promoter polymorphism, which disrupts the binding of ZNF148, is associated with primitive neuroectodermal tumours and ependymomas.

BACKGROUND: Platelet derived growth factor receptor alpha (PDGFRalpha) expression is typical for a variety of brain tumours, while in normal adult brain PDGFRalpha expression is limited to a small number of neural progenitor cells. The molecular mechanisms responsible for the PDGFRalpha expression in tumours are not known, but in the absence of amplification, changes in transcriptional regulation might be an important factor in this process. METHODS AND RESULTS: We have investigated the link between single nucleotide polymorphisms (SNPs) within the PDGFRalpha gene promoter and the occurrence of brain tumours (medulloblastomas, supratentorial primitive neuroectodermal tumours (PNETs), ependymal tumours, astrocytomas, oligodendrogliomas, and mixed gliomas). These SNPs give rise to five different promoter haplotypes named H1 and H2alpha-delta. It is apparent from the haplotype frequency distribution that both PNET (10-fold) and ependymoma (6.5-fold) patient groups display a significant over-representation of the H2delta haplotype. The precise functional role in PDGFRalpha gene transcription for the H2delta haplotype is not known yet, but we can show that the H2delta haplotype specifically disrupts binding of the transcription factor ZNF148 as compared to the other promoter haplotypes. CONCLUSIONS: The specific over-representation of the H2delta haplotype in both patients with PNETs and ependymomas suggests a functional role for the ZNF148/PDGFRalpha pathway in the pathogenesis of these tumours.

Global amplification of mRNA by template-switching PCR: linearity and application to microarray analysis.

Conventional approaches to target labelling for expression microarray analysis typically require relatively large amounts of total RNA, a serious limitation when the sample available is small. Here we explore the cycle-dependent amplification characteristics of Template-Switching PCR and validate its use for microarray target labelling. TS-PCR identifies up to 80% of the differentially expressed genes identified by direct labelling using 30-fold less input RNA for the amplification, with the equivalent of 1000-fold less starting material being used for each hybridisation. Moreover, the sensitivity of microarray experiments is increased considerably, allowing the identification of differentially expressed transcripts below the level of detection using targets prepared by direct labelling. We have also validated the fidelity of amplification and show that the amplified material faithfully represents the starting mRNA population. This method outperforms conventional labelling strategies, not only in terms of sensitivity and the identification of differentially expressed genes, but it is also faster and less labour intensive than other amplification protocols.

Deletion mapping in human renal cell carcinoma.

The highest incidence of renal cell carcinoma (RCC) is reported in Scandinavia. Cytogenetic studies of constitutional tissue in families with hereditary RCC and of sporadic RCC tumor tissue have shown abnormalities of chromosome 3p. In a study of 23 sporadic Scandinavian cases using restriction fragment length polymorphism analysis, we found that 68% of informative patients showed terminal 3p deletions. The break point was not consistent. Loss of a locus on the Y chromosome was seen in 4/14 male patients. Losses of heterozygosity on autosomes included chromosomes 18 (5/15 informative cases) and 17 (3/11 informative cases). Losses in heterozygosity were also found at lower levels for other chromosomes (chromosome 13, 3/16; chromosome 10, 2/19; and chromosome 11, 2/24). The single familial case showed reduplication of part of chromosome 3p and of one chromosome 17. Our data confirm earlier data on losses on chromosome 3p in tumor tissue and by extending this type of analysis to all chromosomes, demonstrate the specificity of this loss. No unique findings were made in the sporadic Scandinavian cases. The results support the thesis that a tumor suppressor gene involved in the oncogenesis of RCC may be located distal to the DNF15S2 locus on chromosome 3p.

Expression of the class VI intermediate filament nestin in human central nervous system tumors.

Tumor cells of a particular tissue may show a pattern of gene expression characteristic of the precursor cells of this tissue. To test this proposition for tumors of the central nervous system (CNS) we have used immunohistochemistry to analyze the expression of nestin in primary human CNS tumors and corresponding nonneoplastic brain tissue. Nestin defines a recently discovered sixth class of intermediate filament proteins and in the rat is expressed predominantly in CNS stem cells. In the adult nonneoplastic human brain we have detected only nestin expression in occasional endothelial cells. In contrast, a variety of primary CNS tumors contained substantially elevated nestin levels. The nestin-positive cells in the tumor tissue were tumor cells and/or endothelial cells. Glioblastomas expressed higher nestin levels than less malignant gliomas. This may indicate a correlation between nestin expression and malignancy within the glioma tumor group. In the primitive neuroectodermal class of tumors we observed both nestin-expressing and nonexpressing tumors, suggesting that nestin expression could be used to further characterize this complex and heterogeneous tumor group. Nine metastatic carcinomas were studied, and none showed nestin immunoreactivity in tumor cells. In conclusion, our data support the notion that primary CNS tumors share gene expression patterns with primitive, undifferentiated CNS cells and that nestin, like other intermediate filaments, may be useful in tumor diagnosis.

Amplification at 12q13-14 in human malignant gliomas is frequently accompanied by loss of heterozygosity at loci proximal and distal to the amplification site.

We have recently reported that a subset of human malignant gliomas shows amplification and overexpression of multiple genes from chromosomal segment 12q13-14, including CDK4, SAS, and MDM2. In the present study we have performed an allelotyping for 16 polymorphic loci spanning both arms of chromosome 12 in a series of 136 gliomas. Allelic deletions were found in 50% (7 of 14) of the malignant gliomas with 12q13-14 amplification and involved loci located on 12q proximal and distal to the amplification site. In contrast, the incidence of allelic loss on chromosome 12 was significantly lower in gliomas without 12q13-14 amplification [14% (11 of 79) in the WHO grade III and IV gliomas, 9% (4 of 43) in the WHO grade I and II gliomas]. The frequent association between 12q13-14 amplification and loss of alleles from 12q is in line with a model suggesting chromosome breakage and deletion as important events in the development of gene amplification.

Delivery of aclacinomycin A to human glioma cells in vitro by the low-density lipoprotein pathway.

The present study shows that a human malignant glioma cell line (U-251 MG) accumulates and degrades low-density lipoprotein (LDL) by a saturable, high-affinity process (Km approximately equal to 5 micrograms/ml). Accumulation and degradation could be enhanced by preincubating the cells in a lipoprotein-deficient medium. The LDL degradation rate was highest when the cells were proliferating rapidly. An aclacinomycin A:LDL complex containing 150 to 450 drug molecules per LDL particle could be obtained by incubating LDL with a large excess of aclacinomycin A at 40 degrees. When the glioma cells were incubated with the aclacinomycin A:LDL complex, cellular drug accumulation was dependent on the LDL receptor activity. There are four reasons for drawing this conclusion. (a) U-251 MG cells with high LDL receptor activity accumulated more drug than U-251 MG cells with low LDL receptor activity. (b) U-251 MG cells accumulated more drug than a mutant fibroblast line (GM 1915) lacking LDL receptor activity. (c) Aclacinomycin A accumulation was increased when U-251 MG cells were incubated in the presence of chloroquine, an agent that inhibits LDL degradation. (d) Aclacinomycin A accumulation was reduced when U-251 MG cells were incubated in the presence of either an excess of native LDL or heparin, which has been demonstrated to inhibit receptor-mediated binding and degradation of LDL. The aclacinomycin A:LDL complex also inhibited growth of the glioma cells. Our results suggest that the glioma cells studied have LDL receptors and that it may be possible to use LDL as a vehicle for lipophilic antineoplastic drugs in order to increase the drug accumulation in tumor cell populations with high LDL receptor activity.

Molecular genetic analysis of chromosome 22 in 81 cases of meningioma.

Constitutional and tumor tissue genotypes from 81 unrelated patients with meningioma were compared at 25 polymorphic loci (restriction fragments length alleles) on chromosome 22. Thirty tumors (37%) retained the constitutional genotype along chromosome 22, a finding consistent with no detectable aberrations on chromosome 22 as studied. Forty-two tumors (52%) showed loss of one allele at all informative loci consistent with monosomy 22 in the tumor DNA. The remaining 9 tumors (11%) showed retained constitutional heterozygosity in the tumor DNA at one or more centromeric loci and loss of the heterozygosity at other telomeric loci, which is consistent with variable terminal deletions of one chromosome 22q in the tumor DNA. The localization of breakpoints in these 9 cases with deletions suggests that a meningioma locus is localized distal to myoglobin locus, within 22q12.3-qter. The male cases showed a higher percentage of tumors with no detectable aberrations on chromosome 22, a finding which may suggest that tumors of males have preferentially smaller rearrangements on chromosome 22q than those of females or that the male and female cases with no detected aberrations have another mechanism of oncogenesis. In view of the recent findings on the localization of the neurofibromatosis-2 gene on chromosome 22, the data from case 11 of our series suggests that the meningioma and the neurofibromatosis-2 loci are separate entities.

Structure and expression of the c-sis gene and its relationship to sporadic meningiomas.

A deletion in an Alu repetitive sequence in the fifth intron of the c-sis gene of meningioma patients was previously described (M. Smidt et al., J. Clin. Invest., 86:1151-1157, 1990). The authors analyzed the structure of this intron in DNA from peripheral blood leukocytes and tumor samples of 86 patients with sporadic meningiomas. After amplifying these DNA sequences by the polymerase chain amplification reaction, the authors failed to find any cases with deletions. They also analyzed the effects on the expression of c-sis of the fifth intron with or without the deletion. A c-sis expression clone with an SV40 promoter was modified by adding introns 4, 5, and 6, and the resulting clones were used to examine the expression of c-sis mRNA in A172, NIH3T3, and Cos-7 cells. Northern blots showed that the quantity of message was not changed when the introns were present and that the size of the message was not changed by the deletion in the fifth intron. The effect of the fifth intron Alu sequence on the c-sis promoter was also tested using clones with chloramphenicol acetyltransferase as a reporter gene in A172 and Cos-7 cells. The c-sis promoter was not affected by the fifth intron Alu sequence with or without the deletion and in either orientation. There were also no effects when cells were stimulated by phorbol 12-myristate 13-acetate or the regulatory gene Tax from human T-lymphotropic virus type 1. These data do not support a role for deletions in the fifth c-sis intron in the development of most sporadic meningiomas.

Amplification of multiple genes from chromosomal region 12q13-14 in human malignant gliomas: preliminary mapping of the amplicons shows preferential involvement of CDK4, SAS, and MDM2.

We have investigated 234 tumors of the central nervous system for amplification of 9 different loci from 12q13-14 and report that about 15% of the anaplastic astrocytomas and glioblastomas show amplification at this chromosomal region. The genes most frequently amplified were CDK4 and SAS (18 of 19 cases). MDM2 was coamplified with CDK4 and SAS in 11 tumors while one glioblastoma showed only MDM2 amplification. Some amplicons additionally included GADD153 (9 cases), GLI (6 cases), A2MR (3 cases), and the anonymous locus D12S8 (2 cases). Either MDM2 or CDK4 and SAS showed the highest amplification level in each individual amplicon and amplification of these genes was consistently accompanied by strong overexpression. Our results thus suggest CDK4, SAS, and MDM2 as main targets for the amplification; however, the possibility exists that all amplicons share a common amplified region between MDM2 and CDK4/SAS which might contain one or more as yet unidentified genes.

CDKN2 (p16/MTS1) gene deletion or CDK4 amplification occurs in the majority of glioblastomas.

Forty-six glioblastomas, 16 anaplastic astrocytomas, and 8 astrocytomas were studied for the loss of the CDKN2 (p16/MTS1) gene on 9p. The CDKN2 locus was homozygously deleted in 19 of 46 glioblastomas (41%) and 1 allele was lost in an additional 13 cases (28%). The deleted regions were limited centromerically in some cases by the MTS2 locus and telomerically by the 1063.7 locus. CDKN2 was homozygously deleted in 3 of 16 anaplastic astrocytomas (19%) and 2 further cases showed loss of 1 allele. Amplification of the CDK4 gene was present in 7 of 14 (50%) glioblastomas and 3 of 11 (27%) anaplastic astrocytomas with no losses at the CDKN2 locus as well as in 2 of 32 (6%) glioblastomas with CDKN2 losses. Thus one or more of these two genes were shown to be aberrant in 85% of glioblastomas and 50% of anaplastic astrocytomas. None of the 8 astrocytomas showed abnormalities of these genes.

Low density lipoprotein receptor activity in human intracranial tumors and its relation to the cholesterol requirement.

The receptor binding of low density lipoprotein (LDL) was determined in homogenates of surgically removed specimens from primary and metastatic intracranial tumors and in some cases also from surrounding brain. Seventy-one specimens from 63 patients were analyzed. In a subsample of 16 specimens from 13 patients, the activity of 3-hydroxy-3-methylglutaryl-CoA reductase was assayed in parallel. The LDL binding in the tumors varied over a wide range. A significantly higher LDL binding activity was found when all tumor samples were compared to brain (P less than 0.05). In the three patients where LDL receptor and 3-hydroxy-3-methylglutaryl-CoA reductase activities were assayed in both tumor tissue and surrounding brain, it was found that the receptor or the enzyme activity was increased in the tumors. It is suggested that certain intracranial tumors have an increased cholesterol requirement and that this may be fulfilled by an enhanced LDL receptor activity or an increased 3-hydroxy-3-methylglutaryl-CoA reductase activity. The data indicate that the LDL receptor activity may be regulated independently of the reductase in intracranial tumors.

Amplification and overexpression of the MDM2 gene in a subset of human malignant gliomas without p53 mutations.

The MDM2 (murine double minute 2) gene has recently been shown to code for a cellular protein that can complex the p53 tumor suppressor gene product and inhibit its function. We studied a series of 157 primary brain tumors and report here that the MDM2 gene is amplified and overexpressed in 8-10% of glioblastomas and anaplastic astrocytomas. Thus, MDM2 represents the second most frequently amplified gene after the epidermal growth factor receptor gene in these tumor types. Sequencing of the p53 transcripts in the cases with MDM2 amplification revealed no mutations and restriction fragment length polymorphism analysis showed, with one exception, no losses of alleles on chromosome 17. Our results indicate that amplification and overexpression of MDM2 may be an alternative molecular mechanism by which a subset of human malignant gliomas escapes from p53-regulated growth control.