Abstract
Background: Schwannomatosis is a recently recognized disorder, defined as multiple pathologically proven schwannomas without vestibular tumors diagnostic of neurofibromatosis 2 (NF2). Some investigators have questioned whether schwannomatosis is merely an attenuated form of NF2.
Methods: The authors identified eight families in which a proband met their diagnostic criteria for schwannomatosis. Archived and prospectively acquired tumor specimens were studied by mutational analysis at the NF2 locus, loss of heterozygosity analysis along chromosome 22, and fluorescent in situ hybridization analysis of NF2 and the more centromeric probe BCR. Linkage analysis could be performed in six of eight families.
Results: Clinical characterization of these kindreds showed that no affected family member harbored a vestibular tumor. Molecular analysis of 28 tumor specimens from 17 affected individuals in these kindreds revealed a pattern of somatic NF2 inactivation incompatible with our current understanding of NF2 as an inherited tumor suppressor gene syndrome. Linkage analysis excluded the NF2 locus in two kindreds, and showed a maximum lod score of 6.60 near the more centromeric marker D22S1174.
Conclusions: Schwannomatosis shows clinical and molecular differences from NF2 and should be considered a third major form of neurofibromatosis. Further work is needed to identify the inherited genetic element responsible for familial schwannomatosis.
Schwannomatosis is a newly defined form of neurofibromatosis (NF) characterized by the development of multiple schwannomas without vestibular tumors diagnostic of NF2. Clinically, patients with schwannomatosis frequently present to the physician with intractable pain rather than neurologic disability, and do not develop other types of nervous system tumors or malignancy. About one third of patients with schwannomatosis are found to have tumors in an anatomically limited distribution, such as a single limb, several contiguous segments of spine, or one half of the body.1,2⇓ Sporadic schwannomatosis appears to be as common as NF23,4⇓ but few cases of familial clustering have been reported. The rarity of familial schwannomatosis stands in stark distinction to NF1 and NF2, which are autosomal dominant diseases with full penetrance.
Tumors from patients with schwannomatosis carry inactivating mutations in the NF2 gene in a spectrum similar to that seen in sporadic and NF2-derived schwannomas.5 Loss of heterozygosity (LOH) for chromosome 22 markers in the region of NF2 is frequently seen with these mutations, implying biallelic inactivation such as that seen in tumor suppressor gene syndromes. Conflicting results have been obtained regarding the presence of germline NF2 mutation in patients with schwannomatosis.6,7⇓ It is not clear if this represents varying diagnostic criteria or genotypic heterogeneity.
In the current study, we report the clinicopathologic features of eight familial schwannomatosis clusterings, and exclude the NF2 region as the germline locus for schwannomatosis.
Subjects and methods.
Subjects.
We studied eight families in which an individual meeting our criteria for schwannomatosis7 gave a history of a relative with one or more schwannomas. All living first-degree relatives of known affected individuals were invited to participate. Pathology reports were requested on all surgical procedures from all affected individuals participating. We have previously described Families 4 and 8, and a portion of Family 5 (Family 4—proband subject 22, Family 8—proband subject 29, and Family 5—proband subject 21).7 To our knowledge, the remainder of the patients have not been reported elsewhere in the literature. This study was approved by the Institutional Review Board of the Massachusetts General Hospital, and informed consent was obtained from all individuals donating tissue for this work.
Specimen collection.
Lymphoblast lines were established from peripheral blood samples as described elsewhere.8 Excess tissue was collected at the time of diagnostic or therapeutic procedures, after pathologic studies were complete. Touch preparations were made by touching the cut edge of the tumor to a sterile glass slide, fixing the slide in 95% ethanol for 2 to 24 hours, and air-drying. High-molecular-weight DNA was extracted from peripheral blood leukocytes, cultured lymphoblasts, and frozen pulverized tumor tissue by sodium dodecyl sulfate–proteinase K digestion followed by phenol and chloroform extractions,7 or by using a Qiagen DNA extraction kit (Qiagen, Valencia, CA).
The majority of tumors studied in this report were collected retrospectively from pathology department archives. Ten 3-μm-thick slices were shaved from each tumor-containing paraffin block. After surrounding paraffin was solubilized with xylenes, the resulting tissue was prepared as previously reported7 or extracted using a Qiagen DNA extraction kit.
Mutational analysis of the NF2 gene.
Single-strand conformational polymorphism (SSCP) analysis of tumor specimens was performed as described elsewhere.7 In brief, the first 15 exons of the NF2 gene were amplified from genomic DNA. Products were diluted in formamide containing buffers and separated on nondenaturing polyacrylamide gels. Products giving aberrant mobility of single or double stranded products were reamplified from both tumor and matching blood specimens. Sequencing was performed bidirectionally using a Big Dye sequencing kit (Applied Biosystems, Foster City, CA) or by manual sequencing using a radiolabeled terminator cycle sequencing kit (USB, Cleveland, OH). In the case of complex insertions or deletions, or of alterations near amplification primers, the exact sequence of the change was determined by T vector cloning.
LOH and linkage analysis.
LOH and linkage analysis was initially determined at the NF2 locus itself using two intragenic microsatellite markers (NF2TET and D22S929), the centromeric marker D22S193, and the telomeric markers D22S268 and D22S430. To define the shared region more exactly in Family 10, the marker D22S1748 was developed in cosmid AC004819 approximately 500 kilobasepairs centromeric to marker D22S268. The distance spanned by these six markers is 2.4 megabasepairs. Subsequent analysis was conducted using a panel of 22 markers distributed over the long arm of chromosome 22. Amplification of alleles using primers and protocols available at the Genome Database (http://gdbwww.gdb.org/) in the presence of P33 was followed by separation on 6% polyacrylamide gels. Loss or retention of heterozygosity was determined by visual inspection of autoradiograms. When subtle LOH was detected, blood-tumor pairs were reamplified a minimum of two times, and results reached by consensus between at least two readers.
Simulation studies on the pedigrees using all individuals for whom DNA was available were performed using the SLINK program.9,10⇓ Two-point lod scores between the disease and individual markers were calculated with the MLINK program of the FASTLINK 3.0P software package,11 a faster version of the original LINKAGE package.12 Multipoint parametric linkage analysis was performed using GENEHUNTER version 1.3.13,14⇓ Simulation studies, two-point, and multipoint analyses were performed assuming an autosomal dominant mode of inheritance, with a penetrance of 80% and a disease-gene frequency of 0.001. Haplotypes were constructed by visual inspection of the linked markers, and confirmed using the haplotype function of GENEHUNTER.
Fluorescent in situ hybridization (FISH) analysis.
Dual-color FISH analysis on archived formalin-fixed paraffin-embedded tissue and touch prep specimens was performed as previously reported.15 Two probes were used, one for the BCR region proximal to NF2 at 22q11.23 (Vysis, Downers Grove, IL) and one for the NF2 region at 22q12 (cosmids n3022 and n24f20, UK HGMP Resource Center, http://www.hgmp.mrc.ac.uk). The cosmid cocktail was directly labeled by nick translation rhodamine and paired BCR/NF2 probes were utilized in each experiment. Nuclei were counterstained with DAPI and hybridization results were viewed on an Olympus BX60 microscope with appropriate single, dual, and triple band pass filters. In general, paraffin-embedded specimens have more cells with single signals than touch preps, consistent with the truncation artifact due to incomplete DNA complement in transected nuclei. Nuclei from non-neoplastic paraffin-embedded tissue specimens, cut at identical thickness, were utilized as disomic controls to establish cutoffs of 50% for deletion by FISH (mean percentage of control nuclei with a single signal plus three standard deviations). For touch preps, cutoffs of 25% were used for deletion, based on prior experience with cytologic material.16 In addition to schwannomatosis tumors, we studied two previously characterized vestibular schwannomas from patients with NF2 as controls. Tumor BL566 is from a family segregating a large deletion of the NF2 locus, which carries a somatic frameshifting deletion in exon 13. Tumor BL690 is from an individual who carries an exon two splice site mutation; a somatic exon nine frameshift was also detected.
Results.
Clinical and pathologic characteristics of subjects.
A total of 31 affected individuals were ascertained by history in these eight pedigrees (figure 1). Five of the 31 were deceased and affected according to relatives’ history only with no available pathology reports (Family 1 I-1, Family 3 I-2, Family 5 II-8, Family 10 II-4 and III-2). A single affected member declined to participate (Family 4 I-1). The remaining 25 affected individuals had undergone a total of 101 surgical procedures for resection of one or more tumors. Four subjects had only a single tumor to their knowledge (Family 3 IV-1, Family 5 II-6 and II-7, and Family 8 II-1). Twenty subjects had undergone cranial MRI scan, and none had a vestibular tumor to their knowledge. Nine of 20 MRI reports were available for direct review, and two subjects had enhancing intracranial lesions. Subject III-6 in Family 10 had multiple dural based enhancing masses consistent with meningiomas. Subject II-2 in Family 11 had a 4-mm left 5th cranial nerve tumor consistent with a schwannoma. No evidence of other intracranial tumors was seen. Clinical characteristics of these patients are summarized in table 1. Fifteen subjects met criteria for definite schwannomatosis and three for probable schwannomatosis.7 No subject met the NIH criteria for NF2, but a single individual (III-6 in Family 10) met the more inclusive Manchester and NNFF criteria.17
Figure 1. Pedigrees of families studied in this report. All probands met clinical criteria for definite or presumptive schwannomatosis.7
Table 1. Clinical characteristics of the patients
An additional nine obligate nonexpressing carriers were ascertained by history. Six of these individuals were deceased at the time the study began and history was available only by relatives’ report (Family 4 II-2, Family 5 II-4, Family 9 II-1 and II-3, Family 10 II-1 and II-6). Three obligate carriers directly participated in this study. Individual III-1 in Family 3 was examined by one of the authors (M.M.) at age 48 and found to be neurologically normal. At age 53 she presented with subarachnoid hemorrhage due to ruptured intracranial aneurysm. Multiple cranial imaging studies did not reveal evidence of tumor, and examination at age 54 continued to show no evidence of cranial or spinal tumor. The other two obligate carriers were not directly examined. Individual III-1 in Family 1 (age 76) and individual III-4 in Family 10 (age 77) had no known tumors at ages 76 and 77; neither had undergone imaging studies. Variability within families was not ascertained outside of diagnostic classification.
Pathology reports were available for 53 surgical procedures performed on 21 of the 25 affected individuals studied. Of the 81 tumors described in these reports, 65 were given a pathologic diagnosis of schwannoma (or neurilemmoma), 2 meningioma (both from individual III-6 in Family 10), and 1 ganglion cyst. Thirteen tumors from six subjects were given a diagnosis of neurofibroma; in each of these six individuals multiple other tumors from the same individual had been given a diagnosis of schwannoma. Archived pathologic material was available from 6 of these 13 cases and was reviewed by a single author (D.N.L.). In all six cases, the tumor was reclassified as schwannoma.
Mutational analysis of tumor specimens.
Adequate tumor material was available for analysis at the NF2 locus from 28 tumor specimens from these individuals. Typical truncating mutations were detected in 18 tumors (table 2). Two large inframe deletions of 60 basepairs (in exon one) and 75 basepairs (in exon three) were detected. In the six instances in which multiple tumors were available from the same patient, no two tumors shared the same mutation. Eighteen of 20 mutations occurred in exons one through eight, which encodes the protein 4.1 domain of the NF2 protein. Single mutations were detected in exons 10 and 11, with the exon 11 mutation occurring in the only meningioma analyzed. Surprisingly, no mutations were detected in exons 12, 13, 14, or 15. None of the mutations detected in tumor specimens were detected in the paired blood specimens by SSCP or directed sequencing and no tumor was found to have two mutations in the NF2 gene. Individual III-6 in Family 10 was found to have a transversion in exon 12 (G1128C) at a codon wobble position in both tumors and the blood specimen. This putative polymorphism was not seen in the other four members of Family 10 studied, and has not, to our knowledge, been reported in the literature.
Table 2. Results of exon scanning of the NF2 gene in tumors from the patients
Loss of heterozygosity.
Microsatellite analysis of these 28 tumors at the NF2 locus revealed that all but four showed LOH (86%). In the five instances in which multiple tumors from the same individual showed LOH, the retained allele (and thus the lost allele) was identical in each tumor. Both tumors from individual III-6 in Family 10 showed loss of the wildtype G at position 1128, consistent with this rare polymorphism lying in cis to the retained allele. In five of six families showing LOH in tumors, the retained allele was identical between family members, although the lost allele often differed (see supplementary figure e1 at www.neurology.org). In Family 10, each of the three individuals studied had different retained and lost alleles.
Adequate material was available for LOH analysis of the entire long arm of chromosome 22 in 18 of these 28 tumors (see figure 2). Loss of all informative markers tested was seen in 10 tumors, consistent with monosomy or mitotic nondisjunction. Five tumors showed loss at and distal to NF2, but varying degrees of centromeric retention, consistent with terminal deletion or mitotic recombination. Two tumors from individuals in Family 9 showed retention of all informative markers tested, whereas a single tumor showed loss centromeric to the NF2 region.
Figure 2. Loss of heterozygosity patterns in tumors from these patients. The position of the fluorescent in situ hybridization probe BCR is shown between markers D22S264 and D22S311. Position of markers on the long arm of chromosome 22 was determined using the MapView function of the NCBI Entrez retrieval system (http://www.ncbi.nlm.nih.gov/Entrez/). The 13 markers shown in bold italics were used in subsequent linkage analysis of these families.
Linkage analysis.
Analysis in the 2.4 megabasepair region from D22S193 to D22S430 was consistent with passage of a single allele to all affected family members and to obligate but nonexpressing carriers in Families 1, 4, 5, and 11 (maximum two point lod score for the intronic marker NF2tet 2.5 at theta = 0). In these families, and in Families 3 and 8, a total of 11 tumors with LOH at the NF2 locus all retained the allele derived from the affected or carrier parent.
In two families (9 and 10), affected members did not share an NF2 region haplotype (two point lod scores, Family 9 at NF2tet = −1.41, Family 10 at D22S1748 = −1.76). We thus sought to determine if a proximal or distal region of chromosome 22 was shared in these families and the original four families. Simulation studies performed on the pedigrees generated a maximum potential parametric lod score of 6.70 demonstrating sufficient power to detect linkage. In addition to six markers in the NF2 region, a panel of eight markers centromeric to NF2 and four markers telomeric to NF2 was haplotyped in all affected members of Families 1, 4, 5, 9, 10, and 11. Multipoint analysis using GENEHUNTER in all families revealed a maximum lod score of 6.60 near marker D22S1174 in the proximal portion of chromosome 22 (see supplementary figure e2 at www.neurology.org). Because Family 9 displays unusual patterns of LOH raising the issue of locus heterogeneity, we repeated analysis without Family 9, yielding a multipoint lod score of 5.04 spanning the region D22S420 to S1148.
FISH analysis.
Adequate tumor material for FISH analysis was available from nine tumors from these subjects, in addition to the two control tumors from patients with NF2 (table 3). All tumors from Family 1 and one tumor from Family 5 showed a single signal at both NF2 and BCR, consistent with interstitial/terminal deletion or true monosomy. However, all tumors from Families 10 and 11 and one tumor from Family 5 showed two FISH signals for each probe, despite evidence of LOH, consistent with mitotic recombination or mitotic nondisjunction. Four of these six tumors were studied a second time in a blinded fashion and gave identical results. Expected results were seen in both control samples.
Table 3. Comparison of fluorescent in situ hybridization (FISH) results with LOH studies
Discussion.
Several investigators have speculated that schwannomatosis is an attenuated form of NF2. To clarify this controversy, we performed clinical and genetic studies on kindreds segregating schwannomatosis. Molecular analysis at the NF2 locus revealed a wealth of alterations in schwannomatosis-derived tumors. The tumor suppressor gene hypothesis would thus predict that the first hit in these tumors would also be present in non-neoplastic cells and would be shared within families. However, we found no evidence of NF2 mutations in normal tissue, and no shared germline NF2 locus alterations in these kindreds. To further clarify this unusual observation, we performed linkage analysis and concluded that a schwannomatosis gene or locus exists centromeric to the NF2 gene on chromosome 22. The relationship of this gene or locus to the NF2 gene itself remains unclear. There was no evidence of locus heterogeneity in this small group of families.
The pathology of tumors in these patients was remarkably homogenous, with strong evidence that all nerve sheath tumors were schwannomas. Equal homogeneity has been found in careful pathologic studies of other forms of NF, where nerve sheath tumors are nearly uniformly neurofibromas in NF1 and schwannomas in NF2.18,19⇓ Interestingly, we found four patients in these families who had only a single tumor. This raises the possibility that some patients with sporadic single schwannomas may carry a germline alteration at the schwannomatosis locus, and genetic risk to the next generation. A single patient was found to have meningiomas, and several other instances of patients with schwannomatosis developing meningiomas have been presented in the literature.20-22⇓⇓ Because meningiomas are common tumors in the general population, further work is needed to determine the significance of this association. No other consistent neoplastic lesions were observed in this population.
Nearly all sporadic and NF2-derived schwannomas appear to stem from inactivation of the NF2 tumor suppressor.23-26⇓⇓⇓ Our own work detected grossly truncating mutations in 62 of 80 sporadic schwannomas (77%).5,27⇓ Immunohistochemical results implicated NF2 loss even in tumors lacking evidence of NF2 gene inactivation.28 Perhaps surprisingly, we observed a clustering of schwannomatosis-derived tumor mutations in the protein 4.1 region of the NF2 transcript (exons one through nine), whereas somatic mutations in sporadic schwannomas appear evenly distributed across both the protein 4.1 region and the alpha helical domain (exons 10 through 15).5,27⇓ LOH of markers around the NF2 locus also appears more common in schwannomatosis-derived tumors (86%) than in our studies of sporadic tumors (50.6%).5,27⇓ Further work is need to determine if molecular differences exist between vestibular schwannomas (which comprise over 85% of the tumors studied in previous reports) and nonvestibular schwannomas.
In contradistinction to our results with these schwannomatosis-derived tumors, there is limited evidence that monosomy for chromosome 22 leads to most LOH in sporadic and NF2-related schwannomas. Our own observations and others4,29⇓ suggest that over 75% of sporadic and NF2-related schwannomas with LOH at NF2 have loss of all 22q markers tested. Extrapolation to a single mechanism is difficult, however, because not all tumors were tested or informative in the most centromeric markers, and no markers are available for the acrocentric short arm. In the Wolff et al. study, 22 of 23 tumors with LOH were felt to have a deletion event (full or partial monosomy) based on scanning densitometry of autoradiograms, whereas a single tumor was felt to have two copies of 22 consistent with mitotic nondisjunction.29 Clonal monosomy 22 was seen in 26 of 51 schwannomas after short term culture30,31⇓ but correlative LOH analysis was not performed in these studies. In an alternative approach, comparative genomic hybridization was applied to 25 sporadic and NF2-related schwannomas, with 22q loss detected in 8 (35%).26 LOH studies were performed in only two tumors with loss, leaving open the possibility that some tumors without 22q loss carried regions of isodisomy. Ideally, a single study that compares molecular events (LOH) with chromosome number and aberration is needed to determine the etiology of second allele inactivation in sporadic and NF-related tumors.
Based on the data generated in this study, we would propose two competing hypotheses for the molecular genetic basis of schwannomatosis. The first is that sch is a second tumor suppressor gene that lies near to NF2 on chromosome 22. In this model, schwannoma formation is dependent on four hits (two in the sch tumor suppressor, and two in the linked NF2 tumor suppressor). An inherited inactivating sch mutation then causes the condition of schwannomatosis. Loss of the chromosome trans to the inherited sch mutation accelerates tumorigenesis by eliminating normal copies of both NF2 and sch; a similar synergism has been observed in the p53/NF1 cis mouse model.32,33⇓ However, the very high rate of observed LOH in schwannomatosis tumors suggests a competing model in which a structural element encourages LOH along the trans chromosome. Such a model is supported by our FISH results showing an unusual tendency toward mitotic recombination (MR) or nondisjunction in some familial schwannomatosis tumors. The issue of MR is an especially intriguing one, because there is large individual variation in MR rate, including about 10% of the human population with extremely low levels of baseline MR.34 Such variation might explain the variable expressivity seen in these families, and the lack of expression in some carriers. MR has recently been shown to be the sole mechanistic pathway to adenoma formation in the mouse model of familial adenomatous polyposis, a tumor suppressor gene syndrome with many molecular similarities to NF.35 The observation that in schwannomatosis tumor loss always occurs trans to the inherited allele suggests the primary event may be a structural one such as a recognition site that promotes mitotic pairing or crossing over, rather then an expressed element that would act equally on either allele. Further work is needed to confirm previous studies indicating that MR is not present at significant frequency in sporadic and NF2-related schwannomas, to further characterize LOH patterns in the proximal portion of 22q, and to study greater numbers of sporadic and familial schwannomatosis tumors using these techniques.
Acknowledgments
Supported by a grant from the National Institutes of Health (RO1-3587804) to M.M.
Acknowledgment
The authors thank the many patients, families, and referring physicians who made this work possible, especially Dr. Arie Weinstock, who initially identified probands in Families 5, 9, and 10.
Footnotes
-
Additional material related to this article can be found on the Neurology Web site. Go to www.neurology.org and scroll down the Table of Contents for the June 24 issue to find the title link for this article.
- Received November 24, 2002.
- Accepted March 20, 2003.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.