Loss of Chromosome 22 Alleles in Human Sporadic Spinal Schwannomas

Bertrand Fontaine, MD, PhD, Mark P. Hanson, AU, Jean Paul VonSattel, MD, Robert L. Martuza, MD, and James F. Gusella, PhD

Acoustic neuromas occur either as sporadic solitary tumors in the general population or as inherited bilateral tumors typically in patients with neurofibromatosis type 2. Loss of heterozygosity for markers on the long arm of chromosome 22 has been reported in both instances, and neurofibromatosis type 2 has been genetically linked to a marker on the long arm of this autosome, suggesting that a unique locus on chromosome 22 is implicated in tumorigenesis of both sporadic and inherited acoustic neuromas. To determine whether the locus for neurofibromatosis type 2 might also be responsible for tumorigenesis of those schwannomas distinct from acoustic neuromas in people without neurofi- bromatosis type 2, we studied the DNA content of three sporadic spinal schwannomas. In all three, we found loss of heterozygosity for at least three markers on the long arm of chromosome 22, indicating a partial or total monosomy 22 in the tumor. Our results suggest that a locus on chromosome 22 is responsible for tumorigenesis in schwann cells regardless of their location in the central nervous system, and that some other mechanism (genetic or nongenetic) might account for the relative high proportion of schwannomas developing from the eighth cranial nerve.

Fontaine B, Hanson MP, VonSattel JP, Martuza RL., Gusella JF. Loss of chromosome 22 alleles in human sporadic spinal schwannomas. Ann Neurol 1991;29.183-186

Schwannomas account for 8% of all primary intra- cranial tumors, most originating from the eighth cranial nerve and forming acoustic neuromas [l, 21. Acoustic neuromas are usually unilateral sporadic tumors. Sei- zinger and colleagues {3] demonstrated loss of hetero- zygosity for restriction fragment length polymorphism (RFLP) markers on chromosome 22 in acoustic neuro- mas, sugesting the involvement of chromosome 22 in their tumorigenesis. Acoustic neuromas are also ob- served as inherited tumors in neurofibromatosis type 2 (NF2). This autosomal dominant genetic disorder is characterized by the development of bilateral acoustic neuromas and other tumors of nerve sheath, glia, and meninges 141. In one large family, NF2 was genetically linked to D22S1, a marker located in the middle of the long arm of chromosome 22 [5 ,6] . Moreover, loss of heterozygosity was reported for chromosome 22 al- leles in tumors developing in patients with NF2 f7], suggesting that the NF2 locus might be implicated in both sporadic and inherited acoustic neuromas.

Schwannomas also develop from nerve roots other than the eighth cranial pair, although much less fre- quently. For example, schwannomas are the most fre- quent tumor affecting the spinal cord [l, 2). The ma- jority are sporadic but they may also be observed in patients with NF2 {4}. Sporadic and NF2 schwan-

nomas are histologically indistinguishable El, 21. We now report molecular evidence supporting a common mechanism for tumorigenesis in both spinal and acous- tic schwannomas. We analyzed the D N A content of three sporadic spinal schwannomas, and in all three we found loss of heterozygosity for at least three markers on chromosome 22, indicating a partial or total mono- somy 22 in the tumor.

Patients and Methods Patients PATIENT 1. A 42-year-old woman developed an asymptom- atic paraspinous mass. A T10-11 Iaminectomy was per- formed and the tumor surgically removed. Histological ex- amination of the tumor revealed a large schwannoma. No clinical evidence of a family history of NF2 was found.

PATIENT 2. A 51-year-old woman developed a clinical his- tory of soreness and stiffness of the neck, with occipital head- ache and pain irradiating to the arms. A posterior cervical laminectomy (C6-7) was performed and the tumor was sur- gically removed. Histological examination of the tumor re- vealed a schwannoma. No family history of NF2 was evident.

PATIENT 3. A 29-year-old woman developed Brown- SCquard syndrome. Spinal cord decompression was per- formed at the T3 level and histological examination of the

From the Molecular Neurogenetics Laboratory, Neuroscience Cen- ter, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA.

Received Jun 7, 1990, and in revised form Aug 1. Accepted for publication Aug 15, 1990.

Address correspondence to Dr Gusella, Molecular Neurogenetics Laboratory, Neuroscience Center, Massachusetts General Hospital, 13th St, Building 149, Charlestown, MA 02129.

Copyright 0 1991 by the American Neurological Association 183

tumor revealed a malignant schwannoma, grade 213. There was no evidence of other tumors in the patient or family. Radiotherapy had been performed on the tumor before sur- gical removal and the molecular examination. We included the tumor in the present study, but we cannot exclude the possibility that radiotherapy may have contributed to the ob- served DNA loss.

Southern Blot Analysts of the Tumors After the tumor was surgically removed, part of it was sent for histological examination and the rest was quickly frozen in liquid nitrogen. The tumor was ground in liquid nitrogen and DNA was extracted according to standard methods [3]. Normal tissue DNA was obtained from blood leukocytes and prepared according to standard methods 131. Restriction endonuclease digestion and Southern analysis of the DNA were performed as described [3] .

Probes Used in This Study Complete description of the polymorphic probes used in this study and their fine localization on chromosome 22 can be found elsewhere [S, 91. The following probes were used: 22CI-18 (D22S10) PstI-RFLP (allele 1: 2.5 kb; allele 2: 2 kb), Bcrysll (CRYBZ) PstI-RFLP (allele 1: 8.2 kb; allele 2: 4.7 to 3.5 kb), DP22 (D22SI j ) SacI-RFLP (allele 1: 8.2 kb; allele 2: 3.6 kb), pMS3-18 (D22SI) BgAI-RFLP (allele 1: 9.5 kb; allele 2: 6.5 kb), and W21C (D22S28) BgA-RFLP (allele 1: 8 kb; allele 2: 6.6 kb).

Results Histopathological features of the three schwannomas studied are illustrated in Figure 1. In molecular studies of the three tumors, loss of heterozygosity was re- vealed by the loss of one allele in tumor D N A at a locus for which the patient was constitutionally hetero- zygous, as revealed by Southern analysis of blood DNA. The tumor of Patient 3 lost allele 2 for CRYB2 (Fig 2A) and allele 1 for 022S28 (Fig 2B). The tumor of Patient 1 lost allele 1 for CRYB2 (Fig 2C), Patient 3 lost allele 2 for CRYB2 (Fig 2D). The weak remain- ing signal for allele 2 (see Fig 2D) came most likely from connective tissue or blood vessels contaminating the tumor sample {3, lo}.

The Table indicates the remaining alleles in the tu- mor for the informative markers on chromosome 22. Note that at least three markers showed allele loss for each tumor. The high relative distance between the markers indicates that the tumors lost at least most of the long arm of chromosome 22.

Discussion Our report of loss of heterozygosity for chromosome 22 alleles in spinal sporadic schwannomas in patients without NF2 extends the previous observation of loss of chromosome 22 alleles in acoustic neuromas 137 and in schwannomas of patients with NF2 171. In accord with our results, Couturier and colleagues Ell} re-

Fig I . Histological appearunce of three sporadic spinal schwan- nomas. (A) Patient I . Photomicrograph of schannoma with moderate& compact spindle cells (Antoni A tissue). (Hematoxylin undeosin. x 31.3 before27% reduction.) (B) Patient 2. Phntorni- rrogruph of schwunnoma with Antoni B tissue djaient t o Antoni A tissue. (Hematoxylin and eosin, x 200 before 27% reduction.) (0 Patient 3. Photomicrograph of malignant schwannornu. The tumor is highly cellular with a necroticfocus (left upper corner). (Hematoxylin and eosin, x 200 before 27% reduction.)

184 Annals of Neurology Vol 29 No 2 February 1991

Loss of Heterozygosity in Spinal Schwannomas

Fig 2. Loss of heterozygosity of chromosome 22 alleles in sporadic rpinal schwannoma, as denzonstrated by Southern blot analysis of blood (B) and tumor (T) DNA. (A) Patient 3 pmbd with Bcrysl I (locus CRYB2). (B) Patient 3 prabed with W23C (locus D22S28i. (C) Patient 1 probed with Bctysll (locus CRYB2). (0, Patient 2 probed with Bcrysll (locus CRYBZ). Blood and tumor DNA wa.r analyzed as dacribed in Patients and Methods. Bar I = allele 1: bar 2 = allele 2.

cently reported, among a series of acoustic neuromas, one sporadic spinal schwannoma with loss of chromo- some 22 alleles. Thus, all four tumors of this type ex- amined to date have displayed loss of chromosome 22 alleles, consistent with a role for a tumor suppressor gene on this chromosome, probably the NF2 locus. We therefore propose that the NF2 locus on chromo- some 22 is very likely to be the predominant locus responsible for tumorigenesis of schwann cells in the central nervous system, regardless of their location. Thus, there must be some additional factor that ex- plains the preponderance of acoustic neuromas over spinal schwannomas. It is possible that loci on other chromosomes are also involved in the formation of

D22S28" D22Sl D22S15 CRYB2 D22SlO Patient Bglr BglII Sad PstI P5tI

- l b - 1 1 2

1 1 - - 2 1 3 2 1 1 1 ND

"References for the probes used in this study can be found in Parients and Methods. In the table, they are ordered according to their local- ization on the long arm of chromosome 22, D22S28 being the more telomeric one and D22S10, the more centromeric one. By conven- tion, all& 1 designates the allele of higher molecular weight and allele 2, the allele of lower molecular weight. In the table, a dash (-) indicates that the patient was homozygous for the marker exam- ined. The number indicates the remaining allele in the tumor DNA compared to blood DNA when the patient was heterozygous for the marker tested. ND = nondetermined. 'DNA wa5 extracted from blood leukocytes and tumors using stan- dard techniques. DNA was digested using appropriate restriction enzymes for the probes used in th s study as indicated in the table. Southern blot analysis was performed as described in Patients and Methods.

these tumors, but the limited number of spinal schwannomas available for investigation have made the search for other chromosomal rearrangements prob- lematic.

Loss of chromosome 22 alleles has also been re- ported for other tumors of the central nervous system, such as meningioma [lo, 12) and a rhabdoidatypical teratoid tumor of the brain [ 131. Although meningio- mas are frequent in patients with NF2, the meningioma locus might be different from the NF2 locus: Meningi- oma can be inherited in families that do not meet the NF2 criteria [14), and results of molecular studies have suggested that the meningioma locus might be more telomeric on the long arm of chromosome 22 than the NF2 locus [12, 157. The continuous description of genetic changes in tumors [16} might lead to a molecu- lar classification of tumors. In that respect, we propose that loss of alleles of chromosome 22 be considered as a genetic characteristic of both schwannoma and me- ningioma. However, studies based on a larger series are needed to confirm ow observation before any clinical application can be done. In that context, advances in polymerase chain reaction technology, now allowing gene dosage studies [l7}, might facilitate routine anal- ysis of chromosomal content in tumors.

The last implication of our results is to suggest that a similar mechanism is involved in the formation of sporadic and NF2-associated schwannornas. The NF2 locus has been proposed to encode a tumor suppressor gene [ S , 71, analogous to the retinoblastorna gene [lS, 19). According to this model, the two alleles of the NF2 locus need to be inactivated to lead to the forma- tion of schwannomas. The observed loss of heterozy- gosity for chromosome 22 alleles in schwannomas is likely to represent one event of inactivation, the loss

Fontaine et al: Spinal Schwannornas and Chromosome 22 185

of total or part of chromosome 22 by mitotic nondys- junction [ 3 ] . Looking for DNA rearrangements with newly available DNA markers on chromosome 22 [9] in schwannomas should permit the discovery of smaller deletions around the NF2 locus, unmasking other events of inactivation of the NF2 locus and helping to characterize the NF2 gene.

This work was supported by National Institutes of Health grants NS22224, NS24279, NS20012, and NS20025, and by NF Inc. (Mass. Bay Area and the National Neurohbromatosis Foundation). Dr Fontaine acknowledges financial support from the International Brain Research Organization.

The authors thank Drs Bernd R. Seizinger and Guy A. Rouleau for valuable discussions.

References 1.

2.

3.

4.

5.

6.

Zdch KJ. Tumors of nerve shearh cells. In: Brain rumors: their biology and pathology. 3rd ed. New York: Springer, 1986:

Russel DS, Rubinstein LJ, eds. Tumors of the cranial, spinal and peripheral nerve sheaths. In: Pathology of tumors of the nervous system. 5th ed. Baltimore: Wihams and Willuns, 1989:533- 571 Seizinger BR, M w a RL, Gusella JF. Loss of genes on chromo- some 22 in tumorigenesis of human acoustic neuroma. N a m e

M-a RL, Eldridge RE. Neurofibromatosis 2 (bilateral acous- tic neuroma). N Engl J Med 1988;318:684-688 Rouleau GA, Wertelecki W, Haines JH, et al. Genetic linkage of bilateral acoustic neurofibromatosis to a DNA marker on chromosome 22. Nature 1987;329:246-248 Wertelecki W, Rouleau GA, Superneau DW, et al. Neurofi-

344-3 5 6

1986;322 :644-64 7

bromatosis 2: clinical and DNA linkage studies of a large kin- dred. N Engl J Med 1988;319:278-283

7. Seizinger BR, Rouleau G, Ozelius LJ, et al. Common pathoge- netic mechanism for three tumor rypes in bilateral acoustic neu- rofibromatosis. Science 1987;236:317-319

8. Human gene mapping 10: tenth international workshop on hu- man gene mapping. Cytogenet Cell Genet 1989;S 1:1-1148

9. Rouleau GA, Haines JL, Bazanovski A, et al. A genetic linkage map of the long arm of human chromosome 22. Genomics

10. Seizinger BR, De La Monte S, Atkins L, et al. Molecular ap- proach to human meningioma: loss of genes on chromosome 22. Proc Natl Acad Sci USA 1987;84:5414-5423

11. Couturier J, Delattre 0, Kujas M, et al. Assessment of chromo- some 22 anomalies in neurinomas by combined karyorype and RFLP analyses. Cancer Genet Cytogenet 1990;45:55-62

12. Dumanski JP, Carlbom E, Collins VP, Nordenskjold M. Dele- tion mapping of a locus on human chromosome 22 involved in the oncogenesis of meningioma. Proc Natl Acad Sci USA 1987;84:9275-9279

13. Biegel JA, Rorke LB, Emanuel BS. Monosomy 22 in rhabdoid or atypical teratoid tumors of the brain. N Engl J Med 1989;321:906

14. McDowell JR. Familial meningioma. Neurology 1990;40: 3 12-3 14

15. Rouleau GA, Seizinger BR, Werreleclu W, et al. Flanking mark- ers bracket the neurofibromatosis cype 2 (NF2) gene on chro- mosome 22. Am J Hum Genet 1990;46:323-328

16" Ponder B. Gene losses in human tumors. Nature 1988;335:

17. Neubauer A, Neubauer B, Liu E. Polymerase chain reaction based assay to detect allelic loss in human DNA: loss of beta- interferon gene in chronic myelogenous leukemia. Nucleic Acids Res 1990;18:993-998

18. Weinberg RA. Oncogenes, antioncogenes and the molecular basis of multistep carcinogenesis. Cancer Res 1989;49:

19. Knudson AG. Mutation and cancer: statistical study of retino-

1989;4:1-6

400-402

37 13-3721

blastoma. Proc Natl Acad Sci USA 1971;68:820-823

186 Annals of Neurology Vol 29 No 2 February 1991