Pediatric Brain Tumors: Loss of Heterozygosity at 17p and TP53 Gene Mutations

Pediatric Brain Tumors: Loss of Heterozygosity at 17p and TP53 Gene Mutations Carmen Orellana, Miguel Hernandez-Martí, Francisco Martínez, Victoria Castel, Jose María Millán, Jose Andrés Alvarez-Garijo, Felix Prieto, and Lourdes Badía

ABSTRACT: Cytogenetic and molecular analyses of primitive neuroectodermal tumors (PNETs) of the central nervous system (CNS) have demonstrated material losses of 17p, the region that contains the TP53 gene, as the most frequent abnormality. Mutations in the TP53 gene are, however, very rare in these tumors. These findings strongly suggest that another, as yet unidentified, gene on 17p may be involved. We performed a search for loss of heterozygosity (LOH) on 17p by microsatellite markers on 26 childhood CNS tumors as well as TP53 gene mutations (exons 5–8) by single-strand conformational polymorphism analysis on 41 pediatric brain tumor samples of distinct histologic types. LOH was detected in 10 cases: 7 PNET, 2 astrocytomas, and 1 glioblastoma mutliforme. In 4 of the PNETs the losses were limited to more distal markers. On the other hand, TP53 mutations were detected in 6 of 41 samples studied. Our results not only confirm the low penetrance of the TP53 gene on pediatric CNS tumors, but also provide further evidence of a putative tumor suppressor gene distal to TP53, between markers (D17S938, D17S926) and 17pter, specifically taking part in the development of PNET. © Elsevier Science Inc., 1998

INTRODUCTION Primary central nervous system tumors are common childhood neoplasms, accounting for 9–20% of all malignancies in the pediatric group. Astrocytomas and primitive neuroectodermal tumors are the most frequent histologic subtypes. Primitive neuroectodermal tumor (PNET) is a generic term for diverse neoplasms having in common, in addition to a postulated derivation from primitive neuroepithelial precursors, a peak incidence in the early years of life and an aggressive clinical biology. PNETs account for about 20% of pediatric primary central nervous system (CNS) neoplasms [1]. Most of the cells of these tumors are small, anaplastic, and undifferentiated. Cytogenetic studies of these tumors have shown isochromosome 17q as the most common structural abnormality [2, 3], with loss of one copy of 17p. Restriction fragment length polymorphism (RFLP) analyses of this area have

From the Unidad de Genetica (C. O., F. M., J. M. M., F. P., L. B.), Servicio de Anatomía Patológica (M. H.-M.), Unidad de Oncología Pediátrica (V. C.), and Neurocirugía Pediátrica (J. A. A.-G.), Hospital Universitario La Fe, Valencia, Spain. Address correspondence to: Carmen Orellana, Unidad de Genetica, Hospital Universitario La Fe, Avenida de Campanar 21, Valencia 46009, Spain. Received October 21, 1996; accepted July 29, 1997. Cancer Genet Cytogenet 102:93–99 (1998)  Elsevier Science Inc., 1998 655 Avenue of the Americas, New York, NY 10010

shown allelic losses in about 50% of these tumors [4–6]. The common area of loss has been restricted to 17p11.2pter. Because TP53, a tumor suppressor gene located in this area, is known to take part in the development of a wide variety of tumor types, including brain tumors, this gene has been implicated in these tumors [4]. Mutations in the TP53 gene are, however, very rare in PNETs, suggesting that another, as yet unidentified, gene on 17p may be involved in these tumors [7–9]. Cytogenetic studies of pediatric astrocytomas have shown a number of numerical and structural abnormalities, but no single characteristic anomaly has been identified. Molecular studies of these pediatric tumors have shown that, in contrast with adult astrocytomas, chromosome 17p deletions in low-grade astrocytomas are rare [5, 10, 11]. Notwithstanding this finding, 17p deletions in pediatric astrocytomas have been shown to be more frequent than would be expected [12]. To assess the role of this chromosomal area in our series of pediatric CNS tumors, seven microsatellites were typed, and single-strand conformational polymorphism (SSCP) analysis of the TP53 exons 5 to 8 was performed. MATERIALS AND METHODS The population studied consisted of 11 females and 26 males who ranged in age from 1 day to 16 years with a

0165-4608/98/$19.00 PII S0165-4608(97)00343-9

94

C. Orellana et al. Table 2 Frequencies of LOH for each tumor type Histologic subtype

Figure 1 Diagram of the most distal part of 17p with the relative position of the markers used. Numbers between markers are estimated distances in centimorgans.

mean of 7 years; two cases were congenital. One medulloblastoma from a 27-year-old man was also included. All these patients were treated at the Hospital Infantil La Fe in Valencia. Our series consisted of 41 pediatric brain tumor samples from 37 patients: in 28 cases, brain tumor samples (6 of which were paraffin embedded) were obtained during the initial surgical procedure before the administration of chemotherapy or radiotherapy. Nine samples were obtained only from patients in relapse, although in 8 of these cases the patient had not received any chemotherapy or radiotherapy. In 4 cases, both primary and relapsed tumor samples were obtained. The patients with medulloblastoma were grouped into two categories: high risk and standard risk. In those patients with high risk, the tumor was multifocal, there was cephalorrhachidian liquid dissemination, or the surgical resection was incomplete. Fresh tumor samples have been collected since 1992 and frozen at 2808C until DNA extraction. In 25 cases where peripheral blood was available, constitutional DNA also was obtained. Genomic DNA from both fresh tumors and lymphocytes was extracted by standard phenol/chlo-

Tumors studied

Cases with loss

PNET-medulloblastoma (P) Astrocytoma (A) Ependymoma (E) Glioblastoma multiforme (GBM) Ganglioglioma (GG) Schwannoma (S)

12 8 2

7 2 0

1 1 1

1 0 0

Total

25

10

roform procedures [13]. DNA from paraffin-embedded tissues was obtained without phenolization according to a previously described method [14]. To analyze the TP53 gene for mutations in the evolutionarily conserved part, exons 5 through 8 were amplified separately by polymerase chain reaction (PCR) as described elsewhere [9], with minor modifications. Detection of SSCP was performed in the following way: 5 mL of PCR product were mixed with 7 mL of water and 8 mL of loading dye containing 95% formamide, 0.05% of both bromophenol blue and xylene-cyanol, and 20 mM EDTA. Samples were heated at 958C for 5 minutes and immediately put on ice for 5 minutes. Five microliters of this mixture was loaded into the gel. All fragments were subjected to electrophoresis under two conditions: with 0% or 5% glycerol, in nondenaturing gels containing 10–15% polyacrylamide, depending on the fragment size, at room temperature. At the same time, both normal and mutated controls underwent electrophoresis in each gel. Bands were later visualized by silver staining [15]. Samples displaying a variant band, indicative of TP53 mutations, were subsequently analyzed by direct sequencing of the PCR product to determine the underlying sequence variation. This was performed by cycle sequencing with the use of dye-labeled terminators (ABI-PRISM Dye terminator). In addition, seven genetic markers, spanning the area between TP53 locus and 17pter (see Fig. 1), were characterized in tumor and constitutive DNA samples to examine loss of heterozygosity (LOH) at 17p. Information concerning these markers is presented in Table 1. After PCR amplification, the products underwent electrophoresis in nondenaturing 10–15% polyacrylamide gels and were silver stained.

Table 1 17p markers used in this study Name D17S5 D17S926 D17S849 D17S570 D17S1176 D17S938 p53CA

Type

Heterozygous (%)

Number of Alleles

Size (base pairs)

Source

VNTR dinuc. dinuc. dinuc. dinuc. dinuc. dinuc.

81 68 50 86 76 90

14 6 5 7 7 15

170–870 243–263 251–261 110–122 95–109 164–180 103–135

ATCC Généthon [16] Généthon [16] Jones et al. [17] Jones et al. [17] Généthon [16] Carbone et al. [18]

Abbreviations: VNTR, variable number of tandem repeats; dinuc., dinucleotide repeat; ATCC, American Type Culture Collection.

95

Pediatric Brain Tumors: LOH at 17p and TP53 Mutations

Figure 2 Examples of LOH for several markers. The letters and numbers at the top denote the diagnosis (abbreviated as in Table 2) and the case number, respectively. The letters in the next row correspond to the sample type: S for constitutional DNA and T for tumor DNA. In all cases, the allelic losses in tumor sample compared with constitutional sample can be seen. For the marker D17S926 in (E), case A-52 does not show a true loss; it shows a mutated allele having a different size from that of constitutional DNA.

RESULTS Loss of Heterozygosity at 17p Twenty-five pediatric CNS tumors were analyzed for LOH at 17p by comparing constitutional and tumoral DNA alleles for each marker. The distribution of histologic subtypes was: 12 PNETs, 8 astrocytomas, 2 ependymomas, 1 glioblas-

toma multiforme (GBM), 1 ganglioglioma, and 1 Schwannoma. We found LOH at 17p in 10 of these tumors, 7 of which were PNETs (7/12), 2 astrocytomas (2/8), and the last glioblastoma multiforme (GBM) (Table 2). Some examples can be seen in Figure 2. Altogether, 5 of these 10 tumors (3 PNETs, 1 astrocytoma, and the GBM) show LOH for every informative marker em-

96

C. Orellana et al. Table 3 LOH at 17p: Results obtained for the loci examined in each tumor where LOH was detected PNET 22 PNET 24 PNET 26 PNET 28 PNET 44 PNET 56 PNET 67 AST 52 GBM 70 AST 77 D17S5 D17S926 D17S849 D17S570 D17S1176 D17S938 P53CA

L P NI

P P

L P P NI NI P P

L L L NI P P P

L L NI L L L

L L L NI P NI P

L NI NI NI NI L

NI M L P P P M

NI L NI L L NI

NI

L

NI NI NI L

NI L L L

Abbreviations: P, preserved; L, lost; NI, noninformative; M, mutated.

TP53 Gene Mutations We analyzed 37 pediatric CNS tumors for mutations in exons 5 through 8 of the TP53 gene. The histologic subtype distribution was: 18 PNETs, 10 astrocytomas, 6 ependymomas, 1 ganglioglioma, 1 Schwannoma, and 1 GBM. Band

shifts were detected in 6 cases (1 PNET, 3 astrocytomas, 1 ependymoma, and 1 ganglioglioma) following SSCP analysis. Table 4 shows the distribution of band shifts detected for each tumor type, and Table 5 the results of sequence analyses in these cases. Some examples are given in Figure 3. A change was found in exon 6 of sample A-35: a silent CGA to CGG transition in codon 213 that was identified by sequencing. This change is a common polymorphism that has been previously described [18]. Two different changes were detected in A-52. Sequencing detected a CAT to GAT transversion in codon 214 (exon 6), which implies the amino acid change from histidine to aspartic acid, and a CGG to TGG transition in codon 282 (exon 8), originating an arginine to tryptophan change. This suggests that both alleles of the TP53 gene were mutated in this case, meaning that genic function of TP53 has been severely impaired. A change was detected in exon 5 of sample A-77, corresponding to a CGC to CAC transition in codon 175, causing an arginine to histidine change. This is a mutational hotspot associated with methylation of the dinucleotide CpG. This mutation coincides with the loss of the second TP53 allele. Changes were detected at exon 7 of samples P-28, GG-33, and E-36. However, the results from the exon 7 sequencing are uncertain. Every sample was sequenced twice in both directions (forward and reverse), but no conclusive results could be extracted. All these mutations were present exlusively in tumor tissue, none of them being constitutional.

Table 4 Distribution of SSCPs detected for each tumor type in the TP53 gene

Table 5 TP53 mutations detected

ployed (Table 3). In the other 4 PNETs, we detected LOH only for distal markers not including either the TP53 locus, the D17S938, or the D17S1176. Moreover, in 2 of them, only D17S5 was affected by the loss, with the most closely linked markers (D17S926 and D17S849) being preserved. Thus, D17S5 is the only marker of this set undergoing loss of one allele in all of these cases. In one case (PNET 44), although the primary tumor did not show LOH, loss was found in a sample from the secondary tumor (a local recurrence). This patient had received both chemotherapy and radiotherapy after the primary tumor extirpation, suggesting that the alterations found in the second tumor could be a consequence of the treatment. Only 2 of the 7 PNET cases with LOH suffered local recurrence, whereas 3 of the 5 cases without observed LOH relapsed. In one astrocytoma, we found LOH only for the marker D17S849. Two markers, D17S926 and p53CA, showed new alleles in the tumor sample when compared with leukocyte DNA, marker D17S5 was noninformative, and the remainder were preserved. These results may be explained by the mutations found in the TP53 gene, which confer genomic instability to the tumor cells, leading to the possible accumulation of mutations.

Tumors analyzed

Band shifts detected

PNET-medulloblastoma (P) Astrocytoma (A) Ependymoma (E) Ganglioglioma (GG) Schwannoma (S) Glioblastoma multiforme (GBM)

18 10 6 1 1 1

1 4a 1 1 0 0

Total

37

7

Histologic subtype

a

Two of them pertained to the same tumor (A-52).

Diagnosisa

Exon

Codon

Sequence change

77 35

A A

5 6

175 213

52

A

214 282

28 33 36

P GG E

6 8 7 7 7

CGC(Arg)→CAC(His) CGA(Arg)→CGG(Arg), polymorphic CAT(His)→GAT(Asp) CGG(Arg)→TGG(Trp) Not sequenced Not sequenced Not sequenced

Case

a

Histologic subtypes as abbreviated in Table 2.

97

Pediatric Brain Tumors: LOH at 17p and TP53 Mutations

Figure 3 SSCP analyses of pediatric CNS tumors. DNA controls for normal and positive mutations are indicated at the top of respective lanes with N representing normal controls and C representing mutated controls. (A) Lane 1 shows an altered band pattern from A-77 (this tumor has also lost the remaining allele of TP53; lanes 2, 3, and 4 show normal pattern from other tumors. (B) Lane 2 shows a band shift found in A-52. (C) Lanes 1 and 2 show the altered band patterns from tumors GG-33 and P-28, respectively. (D) Lane 1 shows the band shift found in A-52.

DISCUSSION We found LOH for several 17p markers in 10 of 25 pediatric brain tumors. The tumors that showed LOH were 7/12 PNETs, 2/8 astrocytomas, and 1 GBM. One of the two astrocytomas (A-52) displayed a loss for one marker (D17S849) and allelic changes for two other

markers (p53CA and D17S926), as well as two mutations in the TP53 gene. For this patient, the diagnosis was xantoastrocytoma. Two samples were obtained from the same tumor, the first at the time of the diagnosis and the second, which showed anaplastic features, from a relapse 5 months after the initial surgery. Both samples had the same molec-

98 ular alterations at 17p. Mutations in the TP53 gene can confer genomic instability to the tumor cells, which may explain the presence of marker alterations in this tumor. The other astrocytoma (A-77) showed loss for all the markers. A mutation was also found in the preserved allele of the TP53 gene in this tumor. This patient was diagnosed as having an anaplastic astrocytoma of the basal ganglia, which does not respond to either radio- or chemotherapy and, at this time, is continuously progressing. These results suggest that TP53 may be implicated in at least some of the pediatric high-grade astrocytomas or in the transition of low- to high-grade pediatric astrocytomas. The only GBM present in our series showed LOH for the marker p53CA, being noninformative for the rest of the markers. Therefore, it was not possible to define the extension of loss. No TP53 gene mutations were detected in the remaining allele, but the possibility cannot be excluded that a mutation is present in regions of the gene not studied or that the SSCP analyses fail to reveal all mutations. In contrast with the preceding implications, the TP53 gene does not seem to be primarily implicated in PNETs. In our series, we detected losses of variable extension in 17p: three tumors (28, 56, and 67) showed LOH for all the informative markers analyzed; of the other four (22, 24, 26, and 44), which presented loss only for markers distal to TP53, two showed LOH just for D17S5. Because this marker appears in genome data base maps as the most telomeric at 17p, the putative tumor suppressor gene is most probably localized near 17pter, distal to D17S926 and D17S849. However, because marker D17S5 is a VNTR, with very different allele sizes (170–870 pb), the possibility exists that some of these losses could be artifactual. We are currently looking for new markers as near D17S5 as possible to clarify this point. It is worth noting that LOH at 17pter was detected only in the recidiva of case PNET-44, being absent from the primary tumor. This fact can be interpreted in several ways: (1) at least in this case, the primary event does not include alterations at 17p; (2) the tumor initially carried two alterations in the putative tumor suppressor gene and subsequently underwent LOH at 17p; (3) alternatively, a dominant mutation, such as that which can occur in the TP53 gene, was present in the primary tumor. In any case, more than one gene can be hypothesized to be present in this region, and the loss may have led to a worsening of the malignancy. It is noteworthy that this patient received both chemotherapy and radiotherapy after primary tumor resection, which may be the cause of the LOH in the local recurrence. In a previously published paper, Cogen et al. [7] suggested that deletion of 17p signified a negative prognosis at least for patients clinically judged to have a good risk. We cannot support this association, even taking into account our limited series and the short follow-up time of some of the cases. Furthermore, our data suggest just the opposite: only two of the seven PNET cases with LOH suffered local recurrence, whereas three of the five cases without observed LOH relapsed, the probability of this distribution being due to chance is 0.2 (Fisher’s exact test). Classifying the 12 tumors as high or standard risk, we found no difference in recurrence of tumors with or without LOH (data not shown).

C. Orellana et al. Our results, in concordance with other published studies, showed that TP53 mutations are very infrequent in pediatric CNS tumors. We found only six mutations in 35 tumors, two of them pertaining to the same tumor. A-52 showed two mutations, located in exons 6 and 8. Another astrocytoma (A-77) showed TP53 mutation in exon 5 and loss of the remaining allele. In PNET, we found only one mutation in exon 7 of tumor P-28, which has not yet been sequenced. This tumor also showed loss of heterozygosity in all 17p informative markers. Two other mutations were in exon 7 of both one malignant ependymoma and one ganglioglioma; they have not yet been sequenced, but no loss of heterozygosity has been detected in these tumors. In conclusion, our results not only confirm the low penetrance of the TP53 gene in pediatric CNS tumors, but also bring further evidence of a putative tumor suppressor gene distal to TP53, between markers D17S938 and D17S926 and 17pter, involved particularly in the initiation or progression of PNET or both. This work was supported by grant no. FIS 95/1825 from the Fondo de Investigaciones Sanitarias. We thank Beatriz Martinez Delgado of the Fundación Jimenez Díaz for kind delivery of the positive controls for TP53 gene mutations. We are also indebted to Ian Costello for his language assistance.

REFERENCES 1. Finlay JL, Goins SC, Uteg R, Giese WL (1987): Progress in the management of childhood brain tumors. Hematol Oncol Clin N Am 1:753–776. 2. Biegel JA, Rorke LB, Packer RJ, Sutton LN, Schut L, Bonner K, Emannel BS (1989): Isochromosome 17q in primitive neuroectodermal tumors of the central nervous system. Genes Chromosom Cancer 1:139–147. 3. Karnes PS, Tran TN, Cui MY, Raffel C, Gilles FH, Barranger A, Ying KL (1992): Cytogenetic analysis of 39 pediatric central nervous system tumors. Cancer Genet Cytogenet 59:12– 19. 4. Cogen PH, Daneshvar L, Metzger AK, Edwards MSB (1990): Deletion mapping of the medulloblastoma locus on chromosome 17p. Genomics 8:279–285. 5. James CD, He J, Carlbom E, Mikkelsen T, Ridderheim P, Cavenee WK, Collins P (1990): Loss of genetic information in central nervous system tumors common to children and young adults. Genes Chromosom Cancer 2:94–102. 6. Thomas GA, Raffel C (1991): Loss of heterozygosity on 6q, 16q and 17p in human central nervous system primitive neuroectodermal tumors. Cancer Res 51:639–643. 7. Cogen PH, Daneshvar L, Metzger AK, Duyk G, Edwards MSB, Sheffields VC (1992): Involvement of multiple chromosome 17p loss in medulloblastoma tumorigenesis. Am J Hum Genet 50:584–589. 8. Biegel JA, Burk CD, Barr FG, Emanuel BS (1992): Evidence for a 17p tumor related locus distinct from p53 in pediatric primitive neuroectodermal tumors. Cancer Res 52:3391– 3395. 9. Badiali M, Iolascon A, Loda M, Scheithauer BW, Basso G, Trentini GP, Giangaspero F (1993): p53 gene mutations in medulloblastoma. Diagn Mol Pathol 2:23–28. 10. Von Deimling A, Louis DN, Menon AG, von Ammon K, Peterson I, Ellison D, Wiestler OD, Seizinger BR (1993): Deletions on the long arm of chromosome 17 in pilocytic astrocytoma. Acta Neurpathol 86:81–85.

Pediatric Brain Tumors: LOH at 17p and TP53 Mutations 11. Rasheed BKR, McLendon RE, Herndon JE, Friedman HS, Friedman AH, Bigner DD, Bigner SH (1994): Alterations of the TP53 gene in human gliomas. Cancer Res 54:1324–1330. 12. Willwert JR, Daneshvar L, Sheffield VC, Cohen PH (1995): Deletion of chromosome arm 17p DNA sequences in pediatric high-grade and juvenile pilocytic astrocytomas. Genes Chromosom Cancer 12:165–172. 13. Maniatis T, Fritsch EF, Sambrook J (1982): Molecular cloning: A laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor. 14. Onadim Z, Cowell JK (1991): Application of PCR amplification of DNA from paraffin embedded tissue sections to linkage analysis in familial retinoblastoma. J Med Genet 28:312–316.

99 15. Biedler JL, Hillard PR, Rill RL (1982): Ultrasensitive staining of nucleic acids with silver. Anal Biochem 126:374–380. 16. Gyapay G, Morissette J, Vignal A, Dib C, Fizames C, Millasseau P, Marc S, Bernardi G, Lathrop M, Weissenbach J (1994): The 1993–1994 Généthon human genetic linkage map. Nat Genet 7, special issue. 17. Jones MH, Sato T, Saito H, Tanigami A, Nakamura Y (1994): Six microsatellite polymorphisms at candidate and confirmed tumour suppressor gene loci. Hum Mol Genet 10:1911. 18. Carbone D, Chiba J, Mitsucloni T (1991): Polymorphism at codon 213 within the p53 gene. Oncogene 6:1691–1692.