SEARCH

SEARCH BY CITATION

Keywords:

  • ependymal tumor;
  • loss of heterozygosity;
  • mutation;
  • neurofibromatosis 2;
  • multiple endocrine neoplasia type 1

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Ependymomas arise from the ependymal cells at different locations throughout the brain and spinal cord. These tumors have a broad age distribution with a range from less than 1 year to more than 80 years. In some intramedullary spinal ependymomas, mutations in the neurofibromatosis 2 (NF2) gene and loss of heterozygosity (LOH) on chromosome arm 22q have been described. Cytogenetic studies have also identified alterations involving chromosome arm 11q, including rearrangements at 11q13, in ependymomas. We analyzed 21 intramedullary spinal, 14 ventricular, 11 filum terminale and 6 intracerebral ependymomas for mutations in the MEN1 gene, which is located at 11q13, and mutations in the NF2 gene, which is located at 22q12, as well as for LOH on 11q and 22q. NF2 mutations were found in 6 tumors, all of which were intramedullary spinal and all of which displayed LOH 22q. Allelic loss on 22q was found in 20 cases and was significantly more frequent in intramedullary spinal ependymomas than in tumors in other locations. LOH 11q was found in 7 patients and exhibited a highly significant inverse association with LOH 22q (p<0.001). A hemizygous MEN1 mutation was identified in 3 tumors, all of which were recurrences from the same patient. Interestingly, the initial tumor corresponded to WHO grade II and displayed LOH 11q but not yet a MEN1 mutation. In 2 subsequent recurrences, the tumor had progressed to anaplastic ependymoma (WHO grade III) and exhibited a nonsense mutation in exon 10 of MEN1 (W471X) in conjunction with LOH 11q. This suggests that loss of wild-type MEN1 may be involved in the malignant progression of a subset of ependymomas. To conclude, our findings provide evidence for different genetic pathways involved in ependymoma formation and progression, which may allow to define genetically and clinically distinct tumor entities. © 2001 Wiley-Liss, Inc.

Ependymomas are glial tumors that arise from the ependymal lining of the ventricular cavities of the brain or the obliterated central canal of the spinal cord. They account for approximately 3–9% of all neuroepithelial tumors and represent the third most frequent brain tumor in children.1 While nearly 90% of pediatric ependymomas are located intracranially, more than 60% of ependymomas in adults arise in the spinal cord or filum terminale.1 The World Health Organization (WHO) classification distinguishes 4 major types of ependymal tumors: the myxopapillary ependymoma (WHO grade I), the subependymoma (WHO grade I), the ependymoma (WHO grade II) and the anaplastic ependymoma (WHO grade III).2 Besides WHO grade, several additional parameters determine clinical outcome of ependymoma patients, including extent of tumor resection, patient age and tumor location. Adult patients with ependymoma usually have a more favorable prognosis than children with histopathologically comparable tumors,3 and spinal ependymomas are generally associated with a better prognosis than intracranial ones of the same WHO grade.4

Most ependymomas exhibit cytogenetic changes. These changes are complex and affect a variety of chromosomal locations. In approximately 30% of ependymal tumors, abnormalities of chromosome 22 have been detected, mainly monosomy.5–11 In addition, loss of chromosomal material has been described for chromosomes 6, 9, 10, 11, 13 and 17, while gains have been reported for chromosome 7.1 Recently, it has been shown that the NF2 tumor suppressor gene, which maps to chromosomal region 22q12, is affected by mutations in a substantial fraction of ependymomas; however, mutations were only detected in those tumors located within the spinal cord.12 In several ependymomas with NF2 mutations, loss of heterozygosity (LOH) involving the long arm of chromosome 22 was also found, indicating complete loss of wild-type NF2 in these tumors.12

Of the various additional cytogenetic changes that have been reported, the affection of chromosome 11 in ependymomas is of particular interest. Several observations suggest that the multiple endocrine neoplasia type 1 (MEN1) gene, a putative tumor suppressor gene that maps to chromosomal region 11q13, could be involved in the pathogenesis of ependymal tumors. MEN1 is a familial cancer syndrome in which affected individuals are predisposed to the development of multiple tumors, mainly of the parathyroids, pancreas and anterior pituitary. Several reports have also described the occurrence of ependymoma in association with MEN1,13–15 including one study in which allelotyping demonstrated LOH that involved the wild-type allele in a thoracic spinal ependymoma.15 Moreover, 6 sporadic ependymoma cases, 3 of which were pediatric, have been reported with intratumoral monosomy 11.5, 8, 16–19 In addition, 4 pediatric cases have shown rearrangements involving 11q13, 3 of which were translocations.9, 20, 21 and 1 was an inversion.5

These findings suggest that loss of MEN1 gene function as well as loss of wild-type NF2 could be involved in the pathogenesis of a subset of ependymomas. In the present study, we have therefore examined a series of 52 ependymomas for mutations in the MEN1 and NF2 genes, as well as for allelic losses on chromosomal arms 11q and 22q.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Tumor and blood specimens, DNA and RNA extraction

Tumor tissue and corresponding blood samples were collected from patients treated at the Department of Neurosurgery, University Hospital Eppendorf, Hamburg between 1991 and 1999. Specimens were stored at −80°C. Genomic DNA from tumor specimens and blood was extracted by using the QIAamp Tissue Kit or QIAamp Blood Kit (Qiagen, Hilden, Germany), respectively, according to the manufacturer′s instructions. To assure that the tumor pieces taken for molecular genetic analysis contained a sufficient proportion of tumor cells, histological evaluation of a representative part of each of these pieces was performed. Only samples with a tumor cell content of more than 80% were included in this study. All tumors were graded according to current WHO criteria.2

Microsatellite analysis for loss of heterozygosity

Primers labeled with fluorescent dyes 6-FAM, or JOE were obtained from MWG-Biotech AG (Ebersberg, Germany). Primer pairs were specific for the following loci: D11S956 (11q13.1), PYGM (11q13.1), D11S987 (11q13.3), INT2 (11q13.3), D22S193 (22q12), NF2CA3 (22q12), D22S268 (22q12) and D22S430 (22q12). PCR amplifications were performed using approximately 20 ng of DNA as template in a total volume of 10 μl. Reactions contained 0.25 U Taq-gold polymerase (Perkin Elmer, Foster City, CA), 2.5 pmol of fluorescence-labeled and unlabelled primer, 1× buffer (10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl2, 0.001% w/v gelatin) and dNTPs (200 μM each). PCR reactions were preceded by a 16 min initial denaturation at 94°C, followed by 40 cycles of 1 min at 94°C, 1 min at 55 °C and 1 min at 72°C in a thermal cycler (Perkin Elmer). Four PCR products (1 μl each) of different sizes and/or labeled with different fluorescent dyes were pooled and dialyzed against distilled water for 1 hr, mixed with 12 μl of formamide and 0.5 μl of Genescan-500 ROX standard (Applied Biosystems, Foster City, CA). After denaturing for 2 min at 95°C, the fluorescence-labeled PCR products were analyzed by using an automated ABI 310 genetic analyzer (Applied Biosystems). Allelic loss was scored as described previously.22

Mutation analysis of the MEN1 gene

To analyze the MEN1 gene (GenBank accession: U93237), a set of PCR primers spanning the coding region of the MEN1 gene from exon 2 to exon 10 and adjacent intronic sequences was generated (Table I). PCR reactions for exons 3 to 9 were performed as described above with an annealing temperature of 65°C using 35 cycles. To amplify exons 2 and 10, touch-down PCR was used with annealing temperatures ranging from 70°C to 54°C and a total of 40 cycles. PCR products were directly sequenced unidirectionally by using the BigDye kit and an automated sequencer (ABI 373, Applied Biosystems, Foster City, CA). Mutations were confirmed by sequencing in both directions. In one case not only tumor DNA but also constitutional DNA was analyzed to exclude a germline mutation (case 37).

Table I. Oligonucleotide Primers for Exon Amplification of the MEN1 Gene
DesignationOrientationSequencePCR product size
Ex2-fForwardGGCTCCTCATTGGGGTGCTTG
Ex2-rReverseGGGTTCTGTAAACCATGGAGGG795 bp
Ex3-fForwardCTGAGGTTGGGTCACAGGCTTGG
Ex3-rReverseGAAAATGGAGTCCCTTGGGTGGC267 bp
Ex4-fForwardCATCACTACCTGGCCCCTTTCC
Ex4-rReverseCCCACAGCAAGTCAAGTCTGGC274 bp
Ex5/6-fForwardCCGATAGGCTAAGGACCCGTTC
Ex5/6-rReverseCCTGCCTCAGCCACTGTTAGG330 bp
Ex7-fForwardGCATTTGTGCCAGCAGGGCAGC
Ex7-rReverseCAGTCCTGGACGAGGGTGGTTG281 bp
Ex8-fForwardCATTTTGCAGTGAGGACCCCAC
Ex8-rReverseCATAGCTGTCTGCTCAGGGCC392 bp
Ex9-fForwardGGTACGTCCTGGCTATGGATTG
Ex9-rReverseCAAGCCCGTGGCTGCTGTCAC402 bp
Ex10-fForwardGTCCTGGAGTTCCAGCCACTG
Ex10-rReverseGCCCTGGGTTCTGAGCTGGAG728 bp

Mutation analysis of the NF2 gene

Mutation screening was performed by using a temperature gradient gel electrophoresis (TGGE) analysis system (Biometra, Goettingen, Germany). Exons 1-15 of the NF2 gene (GenBank accession NM 000268) were PCR-amplified as described previously.23 Exons exhibiting altered banding patterns on TGGE were re-amplified, and after purification, PCR products were directly sequenced in both directions using the BigDye kit and an automated sequencer. Specimens exhibiting LOH 22q were analyzed a second time by TGGE after mixing the respective PCR products at an equal ratio with products obtained from normal control DNA to facilitate heteroduplex formation in cases with NF2 mutations.

Statistical analysis

Associations between tumor localization and LOH 11q or LOH 22q, and between LOH 22q and NF2 mutations, as well as between LOH 11q and LOH 22q were analyzed using the Fisher Exact Test. Age differences between groups of patients with different tumor localizations were compared using the 1-way ANOVA and unpaired t-test.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Clinicopathological findings

A total of 52 tumor specimens from 45 patients were analyzed. From 5 patients more than 1 tumor was available; these additional tumors were local recurrences and in 1 case a metastasis (Table II) Thirty patients were male and 15 were female. The mean patient age at ependymoma surgery was 34.4 years with a range from 0.3 years to 66 years. Of the 52 tumors, 21 were intramedullary spinal, 11 originated from the filum terminale, 14 were of ventricular origin and 6 were located intracerebrally (Table II). Patients with intraventricular tumors were significantly younger (x̄ = 25 years) than patients with intramedullary spinal tumors (x̄ = 42 years) (p = 0.012). While the vast majority (84%) of intramedullary spinal and filum terminale ependymomas were histologically benign, corresponding to WHO grade II or I, 75% of the ventricular and intracerebral tumors were classified as anaplastic, WHO grade III (Table II). The differences in tumor grade and location were also reflected in the differences in patient survival. Only one of the patients with intramedullary spinal ependymoma and none of the patients with filum terminale ependymoma had died by the time of this study, whereas more than 50% of the patients with ventricular or intracerebral ependymoma had died (Table II).

Table II. Clinical, Histopathological and Molecular Genetic Findings in Ependymoma Patients
Tumor numberaAge, sexbOperation yearWHO gradecLocalizationdSurvival statuseMicrosatellite analysisfMutation analysisg
11q22qMEN1NF2
  • a

    Numbers from 1 to 45 were assigned to individual patients; multiple tumors analyzed from the same patient are designated as .1, .2, .3 or .4; rec, recurrence; met, metastasis.

  • b

    Age at time of ependymoma surgery; f, female; m, male.

  • c

    All grade I tumors are myxopapillary ependymomas (no subependymomas were included).

  • d

    Intramed, intramedullary spinal; filum, filum terminale; C, cervical; T, thoracic; L, lumbar; S, sacral.

  • e

    Alive or dead by october 1999.

  • f

    LOH, loss of heterozygosity detected for at least one informative microsatellite locus; H, heterozygosity found for at least one informative locus and no LOH.

  • g

    mut, a mutation was identified; na, not analyzed.

144, m1998IIIntramed C2-C3AliveHLOHmut
247, f1994IIIntramed C2-C4AliveHH
3 (rec)27, m1999IIIntramed T2-T3AliveHLOH
464, f1996IIIntramed C2-C7AliveHLOHmut
557, f1997IIIntramed C4-C7AliveHLOH
633, f1997IIIntramed C2-T2AliveHLOHmut
755, f1999IIIntramed T6-T8AliveLOHH
863, m1997IIIntramed T10-L1AliveHLOHna
9.136, m1998IIIIntramed C6-C7DeadHLOHmut
9.2 (rec)37, m1999IIIIntramed C6-C7DeadHLOHnamut
1050, f1999IIIntramed C1-3AliveHH
1135, m1996IIIntramed C2-C6AliveHLOH
1221, f1993IIIIntramed medulla-C5AliveHLOH
1353, m1999IIIntramed T2-T3AliveHLOHna
1431, f1995IIIntramed C4-C7AliveHH
1544, f1997IIIntramed C6AliveHLOH
1629, m1999IIIntramed T2-5AliveLOHHna
1730, m1995IIIntramed C6-C7AliveHLOH
1819, m1992IIIIntramed C4-T3AliveHH
1956, m1995IIIntramed C2-C3AliveHLOH
2049, m1993IIIIntramed C2-C3AliveHLOHmut
2159, m1999IIFilum, L1AliveHHna
2237, f1997IIFilum, L2AliveHH
2334, m1996IIFilum, L1AliveHH
2437, f1994IIFilum T11-L5AliveHH
2518, m1998IIFilum, L1-L2AliveHH
2628, m1994IFilum L1AliveHH
2726, m1998IFIlum, L5-S3AliveHH
2834, f1996IFilum, L2-L4AliveLOHH
2935, m1995IFilum, T12-S2AliveHH
3038, f1997IIFilum, L1-L2AliveHLOH
3130, m1998IIFilum, L4AliveHLOH
3227, f1993III4th ventricleAliveHLOH
3317, m1991IIITrigoneDeadHH
34 (rec)4, m1992III4th ventricleDeadHLOHna
353, m1996II4th ventricleAliveLOHH
3666, m1997II4th ventricleDeadLOHHna
37.150, f1995II4th ventricleDeadLOHLOH
37.2 (rec)51, f1997III4th ventricleDeadLOHLOHmut
37.3 (rec)53, f1998III4th ventricleDeadLOHLOHmut
37.4 (met)53, f1999IIIIntradural L4-L5DeadLOHLOHmut
38.1 (rec)8, m1995III4th ventricleDeadHH
38.2 (rec)8, m1995III4th ventricleDeadHH
392, m1995II4th ventricleAliveHH
40 (rec)9, m1995III4th ventricleAliveHLOH
410.3, m1997III4th ventricleDeadHH
4227, m1992IIITemporalAliveHH
43.127, m1992IIParietalAliveLOHH
43.2 (rec)32, m1998IIICentralAliveLOHH
444, m1993IIIFrontalDeadHH
45.146, m1995IIITemporalDeadHH
45.2 (rec)47, m1997IIITemporalDeadHH

MEN1 gene mutations and LOH 11q

Of 45 tumor samples analyzed for MEN1 mutations, 3 specimens, all of which were derived from the same patient, displayed a MEN1 mutation (tumors 37.2, 37.3 and 37.4; Table II, Fig. 1c). In these tumors, a G>A mutation at nucleotide 7640 created an immediate stop codon in exon 10 (W471X). Interestingly, in this case the tumor had initially corresponded to WHO grade II (Fig. 1a) and had later progressed towards anaplastic ependymoma (Fig. 1b). We did not detect the mutation in the initial tumor, whereas in 2 subsequent local recurrences and one spinal metastasis it was present. Analysis of constitutional DNA from blood leucocytes revealed no germline mutation (Fig. 1c) and the patient exhibited no clinical signs of MEN1. Microsatellite analysis revealed LOH flanking the MEN1 locus already in the initial tumor as well as in all subsequent manifestations (Fig. 1d). In the initial tumor (37.1), however, the allelic loss appeared to be incomplete, as there was substantial reduction in peak height of the smaller allele of D11S956 and D11S987 when compared with the larger allele, but not a complete signal loss. Microscopically this tumor specimen did not contain more non-tumorous cells, such as blood vessels or neural tissue, than the other tumors. Therefore, it is possible that in the initial tumor, the allelic loss was confined to only a fraction of the tumor cells.

thumbnail image

Figure 1. Histological and molecular genetic analysis of case 37. (a) Histologically the initial tumor (37.1) presented as a well-differentiated, moderately cellular ependymoma, corresponding to WHO grade II. (b) Almost 2 years later the tumor had progressed towards anaplastic ependymoma, WHO grade III, exhibiting poor differentiation with high nuclear pleomorphism, numerous mitoses (arrows) and increased cellularity. (a and b, H&E stain, ×135 magnification.) (c) Mutation analysis of constitutional DNA (Blood) as well as of DNA extracted from the initial tumor (37.1) revealed no germline mutation. In the tumor recurrence (37.2) a hemizygous exchange of a single base (G to A) at nucleotide 7640 was detected, which created an immediate stop codon in exon 10, (W471X); the mutation was confirmed by sequence analysis of the antisense strand. (d) Microsatellite analysis of tumor 37.1 revealed allelic imbalance at marker D11S956, which proximally flanks the MEN1 locus, and at D11S987, which distally flanks the MEN1 locus; at both microsatellite loci a substantial peak reduction was observed for the smaller microsatellite allele. The subsequent recurrence (37.2) displayed virtually complete loss of the smaller microsatellite alleles.

Download figure to PowerPoint

LOH in the 11q13 chromosomal region was further observed in 7 other tumors (obtained from 6 patients). Mutation analysis performed on 5 of these tumors detected no MEN1 mutation. Statistically, LOH 11q was neither associated with tumor location nor tumor grade. In 3 patients, the LOH 11q involved only a subset of the microsatellite loci that were analyzed (Table III). In case 7 allelic loss was observed at D11S987 and INT2, while heterozygosity was retained at PYGM and at a polymorphic locus within the MEN1 gene. In tumors 35 as well as in 43.1 and 43.2, LOH was detected at INT2, whereas heterozygosity was retained at D11S956 and at polymorphic loci within the MEN1 gene.

Table III. Partial Deletions on Chromosome Arm 11q
Tumor numberD11S956 11q13.1PYGM 11q13.1Heterozygous polymorphisms in MEN1D11S987 11q13.3INT2 11q13.3
  1. The MEN1 gene maps between PYGM and D11S98724, 25; n.i., not informative; IVS, intervening sequence.

7n.i.HIVS1-16C>GLOHLOH
35Hn.i.7264C>T (D418D)n.i.LOH
43.1Hn.i.IVS1-16C>Gn.i.LOH
43.2Hn.i.IVS1-16C>Gn.i.LOH

Mutation analysis of the MEN1 gene revealed 5 relatively common polymorphisms. These were C>G in intron 1, 16 nucleotides upstream of exon 2 (15%); S145S (AGC>AGT) (3.4%), R171Q (CGG>CAG) (3.4%); D418D (GAC>GAT) (50%); and A541T (GCA>ACA) (1%). The same polymorphisms have been reported by others before at similar frequencies.27–29

NF2 gene mutations and LOH 22q

Of 7 patients with LOH 11q only 1 case also displayed LOH 22q. Comparing cases with intratumoral LOH on either of the two chromosome arms analyzed, we found a highly significant inverse association between LOH 11q and LOH 22q (p < 0.001).

Of 52 specimens analyzed, NF2 mutations were detected in 6 tumors obtained from 5 patients. All of these tumors were intramedullary spinal (Table II). Two mutations created an immediate stop codon (cases 1 and 4) (Table IV). Two other tumors displayed frameshift mutations, one of which was caused by an insertion (tumor 6) and the other by a deletion (tumor 20). In another tumor (9.1) as well as in a recurrence of this tumor (9.2), a deletion of 16 bp was found in intron 9, 19bp downstream of exon 9. In one additional tumor, in which an aberrant band pattern was observed by TGGE analysis, the alteration turned out to be a G>A transition causing a synonymous polymorphism (tumor 5) (Table IV). Three tumors with NF2 mutations were classified as WHO grade II, and 3 tumors (including 2 from the same patient) corresponded to WHO grade III.

Table IV. NF2 Gene Mutations in Intramedullary Spinal Ependymomasa
Tumor numberExonAlterationConsequence
  • a

    Nomenclature follows the recommendations by den Dunnen and Antonarakis (2000).26

14432C>GY144X
46592C>TR198X
53363G>APolymorphism Q121Q
66584_596dupA199fsX206
9.1IVS9+19_+34delUnknown
9.2IVS9+19_+34delUnknown
208681_690delK227fsX247

In all tumors with NF2 mutations these were associated with LOH involving the 22q12 region. We found a significant association between NF2 mutation and LOH 22q (p = 0.013, n = 45 patients). LOH 22q was observed in 20/45 cases (44%) and was significantly more frequent in intramedullary spinal ependymomas (14/20 cases) than in tumors in other locations (6/25 cases) (p = 0.003).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Analysis of the entire coding sequence of the MEN1 gene in 45 ependymomas revealed a somatic nonsense mutation in 3 tumors, all of which were from the same patient. The mutation creates a stop codon in exon 10 and has been described previously as a germline mutation in a large MEN1-family with 36 individuals.30 Members of that family exhibited tumors of the parathyroid, gastrinomas and prolactinomas, whereas the occurrence of ependymoma was not mentioned. The association of the mutation with clinical symptoms in a MEN1-family indicates that the mutation is functionally relevant. Truncation of the MEN1 gene product Menin, which interacts with the transcription factor JunD,31 most likely results in decreased repression of JunD-activated transcription in these cases.

The combined mutation and LOH analysis in case 37 suggests a sequential order of genetic events on chromosome arm 11q: in tumor 37.1 a substantial fraction of the tumor cells acquired LOH 11q as the first hit, followed later by the occurrence of MEN1 mutation as second hit in tumor 37.2. Likewise, in both subsequent tumors (37.3 and 37.4), a hemizygous somatic MEN1 mutation was also detected. Interestingly, the initial tumor was classified as WHO grade II, whereas all subsequent tumors were anaplastic ependymomas; this finding suggests that loss of MEN1 gene function could be associated with tumor progression in a subset of ependymomas.

Allelic loss on 11q was also observed in 7 tumors, in 5 of which mutation analysis was performed and failed to detect a MEN1 mutation. In these cases, LOH 11q could either be due to chromosomal instability causing random deletion of genomic material or alternatively additional tumor suppressor genes on chromosome arm 11q might be involved in the pathogenesis of some ependymomas. In 3 of these tumors the allelic deletion was only partial and involved microsatellite loci that have been mapped distally to the MEN1 gene locus,24, 25 suggesting the presence of a tumor suppressor gene locus telomeric to MEN1. Evidence for a tumor suppressor gene at 11q13 distally to the MEN1 gene comes from 2 deletion mapping studies in endocrine tumors as well as head and neck squamous cell carcinomas.25, 32 Recently, the human DOC-1R (deleted in oral cancer-1-related) gene has been cloned and mapped to the 11q13 region.33 The DOC-1R gene is a putative tumor suppressor gene, which might be involved in the pathogenesis of human malignancies, but has not yet been analyzed in ependymomas.

In contrast to alterations on chromosome arm 22q, LOH 11q was not associated with any particular tumor location. Interestingly, we found a highly significant inverse association between LOH 11q and LOH 22q. This finding suggests that loss of genetic information on either 11q or 22q could represent independent and alternative mechanisms involved in ependymoma pathogenesis.

Of the 20 patients with intramedullary spinal ependymomas, NF2 mutations were found in 5 patients with a total of 6 tumors, including 1 recurrence. No NF2 mutations were found in tumors of other locations. Our detection rate among intramedullary spinal tumors is comparable with that in a previous study by Ebert et al.12 who discovered 6 mutations in 17 intramedullary spinal ependymomas investigated. In contrast to that previous study, in which NF2 mutations were restricted to WHO grade II tumors, we identified mutations also in 2 patients with anaplastic intramedullary spinal ependymoma. In 4 other previous reports,11, 34–36 in which a total of 38 ependymomas had been investigated, only a single NF2 mutation was detected; it occurred in 1 of 4 intramedullary spinal ependymomas that had been included. In a fifth investigation by Birch et al.,37 who analyzed tumor cDNA, five NF2 mutations were found in 7 intramedullary spinal ependymomas investigated. Our findings thus confirm the previously suggested exclusivity of NF2 mutations in ependymomas of intramedullary spinal location.

Allelic loss on 22q was observed in almost half of all cases analyzed and was significantly associated with intraspinal tumor localization as well as with mutations in the NF2 gene. However, in a substantial fraction of tumors with LOH 22q no NF2 mutation could be identified. One possible explanation of this discrepancy is that due to whole exon deletions we could have failed to detect some NF2 mutations by TGGE analysis. Alternatively, random deletion of genomic material may have caused LOH 22q in some cases. A third possibility is that one or more additional tumor suppressor genes could be located on chromosome arm 22q. Support for the latter hypothesis comes from a recent segregation analysis of a family, not affected by NF2, in which 4 cousins developed an ependymoma.38 In this study, the most probable location of an ependymoma-susceptibility gene was chromosomal region 22pter-22q11.2. Likewise, a previous cytogenetic analysis of a pediatric anaplastic ependymoma also detected a translocation breakpoint at 22q11.2.39 Another study identified a more telomeric breakpoint at 22q13.3.17

Several observations suggest that, if existent, a second tumor suppressor gene on 22q with relevance to ependymomas could be of relatively more importance to ventricular tumors than tumors in other locations. First, whereas in NF2 patients ependymomas usually are intramedullary spinal, all tumors in members of the described ependymoma-family were associated with the ventricular system.38, 40 Second, the percentage of LOH 22q in the present study is relatively high in ventricular epedymomas (40%) despite the fact that NF2 mutations have never been found in these tumors.

In conclusion, the present study suggests that loss of wild-type MEN1 may be involved in the progression towards malignancy in a subset of ependymomas. One or more additional tumor suppressor genes distally to MEN1 on 11q may also be relevant to ependymoma formation. Loss of genetic information from 11q is likely to occur independently of alterations on 22q, and both may represent alternative mechanisms in ependymoma pathogenesis. LOH 22q as well as NF2 mutations predominantly occur in intramedullary spinal ependymomas. Whereas intramedullary spinal ependymomas usually have a favorable prognosis and are curable by microneurosurgical resection, the prognosis of patients with intracranial ependymomas is often unfavorable. Thus, differences in the clinical behavior between intramedullary spinal tumors and intracranial tumors are reflected at the genetic level, where alterations on chromosome arm 22q allow to define a genetically distinct tumor entity.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We are grateful to all patients involved in this study for their cooperation. We thank M. Kolster for excellent technical assistance and S. Freist for assistance with the photographic work. We further thank Dr. U. Knappe, Dr. T. Lücke and Dr. J. Fahrbach for help with the collection of the clinical data and some of the blood samples. Our work was supported by grants from the Deutsche Forschungsgemeinschaft (MW: WE 928/2-1), the Erich und Gertrud Roggenbuck-Stiftung (MW, KL), Wilhelm-Sander-Stiftung (LK) and Deutsche Krebshilfe (WH). This article contains major parts of doctoral theses by LL and UH to be submitted to the Fachbereich Medizin, University of Hamburg.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES