Tumor Biology of Vestibular Schwannomas

Abstract

Recent advances in molecular biology have led to a better understanding of the etiology of many cranial base tumors. The ultimate goal of understanding tumorigenesis cellular mechanisms is to develop new treatment options that improve patient outcomes. NF2-related schwannomatosis, formerly known as neurofibromatosis type 2 (NF2), is an inherited genetic disorder associated with bilateral vestibular schwannomas (VS). This chapter discusses the structure and function of the NF2 gene, associated signaling pathways, candidate genes involved in NF2 tumorigenesis, and phenotypic-genotypic correlations. Recent advances in treatment options, including pharmacological and gene therapeutic approaches, are also discussed.

Introduction

Recent advances in molecular biology have led to a better understanding of the etiology of many cranial base tumors. The ultimate goal of understanding the cellular mechanisms of tumorigenesis is to develop new treatment options that improve patient outcomes. Indeed, trials of several therapeutic agents targeting known pathways are underway.

As an example of the recent advances in cranial base tumors, vestibular schwannomas (VS) will be highlighted. VS generally arise as sporadic unilateral tumors in the internal auditory canal (IAC) and the cerebellopontine angle (CPA); however, they are also found bilaterally as the hallmark of the inherited genetic disorder NF2-related schwannomatosis, formerly known as neurofibromatosis type 2 (NF2) [Online Mendelian Inheritance in Man (OMIM) #101000]. NF2 syndrome is an autosomal-dominant disorder caused by the pathogenic biallelic variants on chromosome 22, which gives rise to VS, as well as meningiomas, ependymomas, epiretinal membranes, hamartomas, and presenile cataracts.

Fundamental understanding of the underlying molecular events leading to tumor formation began when the neurofibromatosis type 2 tumor suppressor gene (NF2 gene) was identified, and pathogenic variants were confirmed in the VS or the germ line [1,2,3]. The clinical characteristics of VS and NF2 syndromes have both been related to alterations in the NF2 gene. The clinical manifestations of NF2 have been described in association with various pathogenic variants, and signaling pathways affected by merlin (the product of the NF2 gene) are becoming better understood.

For purposes of this chapter and elsewhere in the literature, abbreviations are as follows: “NF2” is used to indicate the human disease of neurofibromatosis type 2; the italicized “NF2” represents the human neurofibromatosis type 2 gene; and “Nf2” indicates the homolog or similar gene expressed in rodents. The mouse Nf2 gene and the human NF2 gene share greater than 98% amino acid homology, thus making the mouse a relevant model for study. Interestingly, chickens have 94% and drosophila have 50% human amino acid homology as well [4].

The nomenclature nf^+/+ is wild-type or normal nf gene. nf^−/− refers to the absence of both alleles (homozygosity), and nf^+/− refers to heterozygosity or one allele only being nonfunctional.

Vestibular schwannomas can be grouped into unilateral sporadic solid VS or bilateral NF2-associated schwannomas [5]. VS with true cysts or cystic schwannomas present with a unique and aggressive phenotype, which may distinguish it from the more benign solid unilateral sporadic schwannoma. Pathogenic variants in the NF2 gene have been identified in all three tumor types, but at the molecular level, the tumors are only beginning to be distinguished from each other [6, 7].

Schwannomas most commonly present as solitary tumors and constitute 95% of all VS. Sporadic unilateral VS occur in approximately 10–20 persons per million per year [8, 9]. However, the true incidence may be higher. Anderson et al. [10] reported an incidence of 7 unsuspected schwannomas per 10,000 brain MRI studies (0.07%) or 70 per million. Sporadic tumors usually occur in the fourth and fifth decades with a mean presentation at about 50 years of age. Although histologically benign, schwannomas can compress the brain stem, leading to hydrocephalus, herniation, and death. Most commonly, they are associated with hearing loss, tinnitus, imbalance, facial paralysis, and facial paresthesias.

NF2 [OMIM #101000] is clinically an autosomal-dominant disease that is highly penetrant [11]. NF2-associated tumors account for about 5% of all VS. Patients who inherit an abnormal copy of the NF2 tumor suppressor gene have a 95% chance of developing bilateral VS. However, about one-half of patients have no family history of NF2 in Caucasian populations. In a study by Teranishi et al., the number of familial NF2 cases in Japan was found to be around 7% [12]. As founders, patients without family history of NF2 represent new germ line mutations that were not inherited. Other disease features of NF2 include intracranial meningiomas, ependymomas, spinal schwannomas, epiretinal membranes, hamartomas, and presenile lens opacities [13,14,15,16]. The onset of symptoms is usually between 11 and 30 years of age, but for some patients, the symptoms may present in their fifth or sixth decade. The incidence is estimated at between 1 in 33,000 and 1 in 40,000 [17, 18]. NF2 is not to be confused with neurofibromatosis type 1 (NF1), historically referred to as von Recklinghausen’s syndrome [OMIM #162200]. NF1 is associated with multiple peripheral neurofibromas and is caused by a pathogenic variant in the NF1 tumor suppressor gene on chromosome 17.

The spectrum of phenotypic expression in NF2 varies widely. In the past, the more severe clinical presentations were termed the Wishart type [19]. In addition to bilateral VS, patients often suffered from associated spinal tumors with a typical onset of symptoms in the late teens or early 20s. The less severe phenotypes have been termed the Gardner type with a later onset of bilateral schwannomas and fewer associated intracranial or spinal tumors [20]. A third subtype known as segmental or mosaic NF2 has been described, where only a portion of cells carries the NF2 pathogenic variant due to a postzygotic mutation [21]. In truth, the spectrum of disease does not always allow a clear distinction but rather follows a continuum.

Mosaic NF2 may vary in its clinical presentation compared to familial cases of NF2 [22]. The cause of segmental NF2 may be due to somatic mosaicism where a pathogenic variant occurs later in embryogenesis rather than in the germ line DNA; therefore, only a portion of the patient’s cells carry the pathogenic variant, and the disease is manifested in limited areas of the body [23, 24]. In contrast, patients with familial NF2 inherit one pathogenic variant from a parent at conception, and all cells carry one mutant allele. Kluwe et al. [22] estimated that mosaicism may account for 25% of the NF2 cases of any subtype among patients whose parents did not display the disease. Patients with somatic mosaicism could display bilateral VS if the postzygotic mutation occurred early in embryogenesis. However, they may also display an atypical presentation or segmental NF2, in which the patient has a unilateral vestibular schwannoma and an ipsilateral, additional intracranial tumor, such as a meningioma, if the postzygotic mutation occurred later in development [25]. Unlike the traditional forms of NF2, the risk of passing NF2 caused by mosaicism to future offspring is low.

Cystic Schwannomas

Cystic VS are a particularly aggressive group of unilateral, sporadic schwannomas, which invade the surrounding cranial nerves, splaying them throughout the tumors. Cystic VS are associated with either intratumoral or extratumoral cysts, which develop within Antoni B regions, a histologic term to describe loosely organized tissue within peripheral nerve sheath tumors. In addition, a higher degree of nuclear atypia is seen in cystic tumors [26, 27]. Careful distinction must be drawn between the truly cystic schwannomas and the very common heterogeneous schwannomas, which are not as aggressive in their clinical behavior. On magnetic resonance imaging (MRI), truly cystic regions of the tumors are hyperintense on T2-weighted images, and the cysts do not enhance with gadolinium administration. The noncystic component of cystic tumors enhances with gadolinium in a manner similar to the unilateral and NF2-associated schwannomas. Cystic tumors may grow rapidly and are difficult to manage due to the high rate of hearing loss, facial nerve paralysis, and recurrence that occurs after surgical removal [28]. When compared to solid tumors of a similar size, the rate of complete facial nerve paralysis (House-Brackmann grade VI) with surgical removal of cystic tumors was 41%, as compared to 27% for that of solid unilateral schwannomas [29]. Cystic tumors are also more likely to have continued growth and facial nerve paralysis when treated with stereotactic radiation therapy than either the unilateral spontaneous or NF2-associated schwannomas [30].

Although the effectiveness of treatment with current surgical and radiation treatments for VS is generally good, treatment-related morbidity continues to be problematic.

A brief review of recent discoveries and advances in the molecular biology of VS will follow.

The NF2 Gene

The NF2 gene was localized to chromosome 22 band q12 through genetic and physical mapping (Fig. 10.1) [31, 32]. Positional cloning studies led to the discovery of the NF2 gene in 1993 by two independent groups, Rouleau et al. [2] and Trofatter et al. [1] The gene was found to be frequently mutated in NF2-related VS. Since that time, pathogenic variants in the NF2 gene have been found not only in NF2-associated tumors but also in sporadic unilateral schwannomas and cystic schwannomas [3, 33,34,35,36,37,38,39,40]. Additionally, pathogenic variants within the NF2 gene have been frequently identified in meningiomas and occasionally identified in other tumor types such as mesotheliomas [36,37,38, 41,42,43,44,45].

Fig. 10.1

Summary of genetic principles. Pathogenic variants can occur upstream of the gene itself in the promoter or regulatory regions or at the boundaries between introns and exons (splice sites). Pathogenic variants in the promoter or regulatory region may lead to over- or underexpression of the gene, while mutations in the intron/exon boundary may lead to aberrant splicing. Aberrant splicing might delete part of the resulting protein, add a new section of amino acids, or result in a frameshift. Single-nucleotide pathogenic variants within the exon may be classified as missense (single amino acid change), nonsense (premature stop codon), or frameshift (altered reading frame with multiple changes in downstream amino acids). Frameshift and nonsense pathogenic variants can be particularly disruptive to protein function. (Adapted from Surgery of the Ear. 6th Ed. Eds. PMPH. 2010)

Human NF2 Pathogenic Variants and Their Clinical Correlation

Specific NF2 pathogenic variants in VS have been characterized in sporadic unilateral schwannomas and NF2-associated schwannomas [32,33,34,35,36, 38, 46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61]. The frequency, type, and distribution of NF2 pathogenic variants were shown to be different between the sporadic and familial NF2 tumors [3]. Point mutations accounted for the majority of pathogenic variants identified in NF2 patients, whereas small deletions accounted for the majority of pathogenic variants found in sporadic unilateral tumors [47, 48, 56].

Two groups have recently published large databases of NF2 pathogenic variants. Ahronowitz et al. [62] presented a meta-analysis of 1070 small genetic changes detected primarily by exon scanning including 42 intragenic changes of 1 whole exon or larger and 29 whole gene deletions and gross chromosomal rearrangements. Overall, somatic events detected by tumor analysis showed a significantly different genetic profile than constitutional events detected in the patients’ leukocytes. Over half of the pathogenic variants detected in tumors were frameshift pathogenic variants. The deletion of a single nucleotide from a gene can have particularly disrupting effects because all the codons downstream from the mutation are changed. In comparison, constitutional or system-wide inherited changes were primarily nonsense and splice site pathogenic variants (Fig. 10.1) [63]. Somatic events also differed markedly in meningiomas where most pathogenic variants were within the 5-prime protein 4.1 (ezrin, radixin, and moesin) (FERM) domain of the transcript with a complete absence of pathogenic variants in exons 14 and 15. Less than 10% of small alterations were nontruncating or lesions, which would not completely destroy the protein. These changes were clustered in exons 2 and 3, suggesting that this region may be especially crucial to tumor suppressor activity in the protein. We and others have shown a frequent pattern of mutation changing arginine codons (CGA) to the thymine stop codon (TGA). This type of “stop” codon not only terminates the transcription of mRNA but often results in degradation of the mRNA already transcribed in both VS and meningiomas, causing a truncating lesion [3, 64]. The chemical bonds associated with the configuration of these nucleotides make them less stable and more likely to undergo spontaneous mutation not only in NF2 but many other tumors as well. No pathogenic variants have been reported in exons 16 and 17, and few were noted in exon 9.

Studies to determine if genotype could be a predictor of disease severity showed that deletions that caused truncation of the NF2 protein have a more severe phenotype in NF2 pedigrees [47, 48, 56], while missense mutations (Fig. 10.1) or small in-frame insertions in the NF2-coding region have been reported to associate with a mild phenotype [3, 36, 37, 55, 57, 58]. However, exceptions are common as some missense mutations have been associated with severe phenotypes. The location of the mutation within the gene may influence phenotype as missense mutations within the α-helical domain of the NF2 protein are associated with a less severe phenotype than those within the conserved ERM domain [62]. This lack of genotype-phenotype correlation was also seen for large deletions, which could give rise to mild phenotypes as well as the previously reported severe disease expression [65].

Recent studies have suggested that genetic severity scores of NF2 may have direct clinical implications, such as stratification during clinical trials. Phenotypic scales, such as the NF2 impact on quality of life (NFTI-QOL) survey, have been shown to be moderately associated with genetic severity. One of such genetic severity scales was validated on a cohort of 142 patients and has been proposed with the following genotype-phenotype correlations:

• Type 1 cases present with a mild phenotype and include mosaic cases where the pathogenic variants are seen only in the tumor and not in the blood.

• Type 2a cases present with a mild phenotype and include missense mutations, truncating mutations in exons 1 and 13/14, splicing variants in exons 7–15, and mosaic variants in the blood (excluding those in exons 2–13).

• Type 2b cases present with a moderately severe phenotype and include large deletions, splicing variants in exons 1–6, and mosaics for truncating variants (exons 2–13) in the blood.

• Type 3 cases present with a severe phenotype and are frequently childhood onset with an expected mortality before the age of 60 years. These cases involve truncating mutations in exons 2–13.

The genetic severity score has been found to correlate with various hearing outcomes, including pure tone average (PTA) deterioration, hearing classification, and optimum discrimination scores (ODS). A weak correlation has also been reported between quality of life outcomes and the number of interventions. Genetic severity scores for NF2 have been used to counsel patients and may be implemented in clinical trial design, such as confining early phase trials to type 2b and 3 cases [66, 67].

This classification system has been supplemented with additional genotypic and phenotypic data. A cohort of 57 NF2 patients was sequenced and found to have a variety of germ line NF2 mutations, including truncating mutations, large deletions, splice site variants, and missense mutations, as well as mosaic NF2 mutations. Patients were assessed for progression of functional disability in hearing, swallowing, and gait function. The length of functional preservation differed significantly by mutation type, with longer functional preservation in mosaic variants and shortest preservation in truncating mutations. The authors used their findings and the previously described genetic severity score to generate a revised scale. Cases were scored by genotype: 0 for mosaic and variants in exons 1, 14, and 15; 1 for missense mutations, large deletions in exon 1 or promoter, splice site variants in exons 8–15, and mosaic variants in exons 2–13; 2 for large deletions excluding exon 1 or promoter and splice site variants in exons 1–17; and 3 for truncating variants in exons 2–15 and classified into mild (0, 1), moderate (2), and severe (3). Age of onset was incorporated into the genetic severity score, where cases scored as 0 had an age of onset at or after 25 years and cases scored 1–3 had an age of onset before 25 years. NF2 class predicted various clinical outcomes, including length of functional preservation, progression of functional disability, the schwannoma growth rate, and the number of interventions per year. As more cohorts are studied, genotype-phenotype correlation analyses may help predict functional outcomes and inform long-term disease management for patients with NF2 [12].

Clinical studies indicate that the phenotypic expression is more closely related within families than between families, but even within families, variability exists [68]. When statistically analyzed, the age of onset of NF2, the age at onset of hearing loss, and the number of intracranial meningiomas were all found to have a significant intrafamilial correlation [69]. It is not known whether this variability occurs by chance, by epigenetic phenomena, or by modifying genes at other loci [70]. Epigenetic phenomena are not caused by a change in the coding sequence of the DNA but rather alter the gene expression during gene development or cell proliferation. An example would be the methylation of segments of the promoter region of the NF2 gene, which suppresses the expression of the gene. The NF2-coding region encompasses 17 exons spanning 90 kilobase pairs (kBp) of DNA on chromosome 22 [1, 2, 71]. By extensive screening of the entire NF2 gene, Zucman-Rocci et al. [69] reported an 84% mutation detection rate in VS. Thus, additional mechanisms of inactivation of the NF2 gene in some patients may exist. Mutation or methylation in the regulatory region of the NF2 gene has been suggested (Fig. 10.1) as a possible mechanism of gene inactivation [72,73,74]. Posttranscriptional alternative splicing and differential polyadenylation may also be considered as possible means of inactivating the NF2 gene [71]. Polyadenylation refers to the variable number of adenines that are attached at the 3′ end of the mRNA which may play a role in the way a gene is regulated. Alternative splicing refers to different patterns of splicing exons, in this case, the NF2 gene, resulting in differing isoforms of the gene. The two most common isoforms of NF2 have either exon 16 or exon 17 spliced out.

The NF2 Protein: Structure and Function

The NF2 gene product is named merlin, for moesin-ezrin-radixin-like protein, or schwannomin, a name derived from schwannoma [1, 2]. For simplicity, the NF2 protein is referred to as “merlin” in this chapter. The NF2 gene is transcribed into multiple RNA species by alternative splicing. Isoforms I and II are the two major RNA isoforms expressed in the cell. Isoform I encodes a 595-amino-acid protein and has exon 16 spliced out. Isoform II, containing all 17 exons, differs from Isoform I only at the C-terminus. Insertion of exon 16 into the mRNA provides a new stop codon, resulting in a 590-amino-acid protein that is identical to isoform I over the first 579 residues. Intriguingly, only isoform I possesses growth suppression activity [75].

Merlin shares a high degree of homology to the erythrocyte protein 4.1-related superfamily of proteins, which act to link the actin cytoskeleton to the plasma membrane [1, 2]. In particular, three proteins, ezrin, radixin, and moesin, referred to as the ERM family, share a great deal of structural similarity with merlin [76, 77]. The proteins belonging to this family have a similar N-terminal globular domain, also known as the FERM domain, followed by an α-helical stretch, and finally a charged carboxyl terminus [75, 78, 79]. The key functional domains of merlin may lie within the highly conserved FERM domains and the unique C-terminus of the protein. The ERM proteins have been shown to be involved in cellular remodeling involving the actin cytoskeleton [80]. These proteins bind actin filaments in the cytoskeleton via a conserved C-terminal domain and possibly via a second actin-binding site in the N-terminal half of the protein [81, 82].

Like the ERM proteins, merlin is expressed in a variety of cell types where it localizes to the areas of membrane remodeling, particularly membrane ruffles, although its precise distribution may differ from the ERM proteins expressed in the same cell [83]. Membrane ruffling is related to the interaction between the cytoskeleton with the cell body and the cell membrane. Schwannoma cells from NF2 tumors show dramatic alterations in the actin cytoskeleton and display abnormalities in cell spreading as well [84]. These results suggest that merlin may play an important role in regulating both the actin cytoskeleton-mediated processes and cell proliferation [85]. However, it should be noted that merlin has a growth suppression role, while other ERM family members seem to facilitate cell growth.

Merlin Acts as a Tumor Suppressor

Overexpression of the NF2 gene in mouse fibroblasts or rat schwannoma cells limits cell growth [61, 86] and suppresses cell transformation by the ras oncogene [87]. Without the NF2 gene overexpressed, transformation occurs when a cancer-causing protein, ras, allows the cellular proliferation machinery to become activated without proper regulatory mechanisms in place, causing malignant tumor growth. The growth control of certain Schwann cells and meningeal cells is lost in the absence of NF2 function, suggesting that NF2 mutations and merlin deficiency disrupt some aspect of intracellular signaling that leads to cellular transformation. Together, these findings demonstrate merlin’s ability to act as a growth suppressor.

Nf2 knockout mouse models, which are either heterozygous or homozygous for the Nf2 gene in the germ line, have been created [88, 89]. Heterozygous Nf2 knockout mice go on to develop osteosarcomas and, less often, fibrosarcomas or hepatocellular carcinomas [86]. Metastatic disease is common in this model. Genetic analysis shows that nearly all these malignant tumors are missing both Nf2 alleles. Tumor growth in the absence of both Nf2 alleles indicates that the Nf2 gene possesses a classical tumor suppressor function. However, none of the heterozygous Nf2 mice develop tumors or clinical manifestations associated with human NF2. The Nf2 gene also plays an important role during early embryogenesis. Homozygous Nf2 mutant mice, which are missing both Nf2 alleles, died at approximately 7 days of gestation from a gastrulation defect or failure of inward migration of cells. Merlin is important in cell migration during embryogenesis [90].

By engineering mice whose Schwann cells have exon 2 excised from both Nf2 alleles, conditional homozygous Nf2 knockout mice have been produced, which display some characteristics of NF2 including schwannomas, Schwann cell hyperplasia, cataracts, and osseous metaplasia [91]. A conditional knockout model is engineered by inserting a regulatory mechanism into the gene, allowing the investigator to choose the time at which the desired gene (in this case Nf2) will be knocked out. Although these results are in favor of the argument that loss of merlin is sufficient for schwannoma formation in vivo, none of the tumors observed in these conditional knockout mice were found on the vestibular nerve. This contrasts with those VS commonly found in patients with NF2.

Biallelic NF2 inactivation is also frequently found in sporadic and in NF2-associated meningiomas. By engineering mice whose arachnoidal cells have exon 2 excised from both Nf2 alleles, homozygous Nf2^−/− mice show a range of meningioma subtypes histologically similar to human tumors [92]. Taken together, these results demonstrate a tumor suppressor function for merlin in both Schwann cells and arachnoidal cells, from which meningiomas commonly originate.

Merlin Signaling and Regulation

Merlin overexpression (Fig. 10.2), unlike other members of the ERM protein family, causes growth suppression. In addition to the actin cytoskeleton, merlin has been shown to associate with cell membrane domains, which are highly enriched in signaling molecules that regulate cellular responses to proliferative and antiproliferative stimuli [93]. Vestibular schwannoma cells with NF2 inactivation have dramatic alterations in cell spreading. To date, several proteins that are likely to interact with merlin have been identified including the ERM proteins [94, 95], CD44 [96, 97], F-actin [98], paxillin [99], microtubules [100], BII-spectrin [101], B1-integrin [102], B-fodrin [99], the regulatory cofactor of Na+-H+ exchanger (NHE-RF) [103, 104], SCHIP-1 [105], hepatocyte growth factor-regulated tyrosine kinase substrate (HRS) [106, 107], p21-activated kinase 1 and 2 (Pak1 and Pak2), Rac1, the RIB subunit of protein kinase A [108, 109], components of cadherin-mediated cell junctions [110], PIKE-L [phosphatidylinositol-3-kinase (PI3K) enhancer] [111, 112], erbin (erbB2-interacting protein) [113, 114], Hippo [115, 116], and NF-kB [84, 117, 118]. Major protein interactions are summarized in Table 10.1.

Fig. 10.2

Schematic diagram of merlin action. This diagram shows how Rac1 and Pak help convert the merlin protein from a closed conformation to an open conformation by phosphorylation of the protein. Consequently, merlin, in its open conformation, can interact with CD44 and facilitate linking the actin cytoskeleton to the cell membrane. p21-activated kinase 2 (Pak2) has been shown to phosphorylate merlin at serine 518 and thereby inactivating its function. Merlin is activated by dephosphorylation by myosin phosphatase (MYPT-1-PP1δ). (Adapted from Surgery of the Ear. 6th Ed. Eds. PMPH. 2010)

Table 10.1 Major protein-protein interactions that modulate merlin’s growth regulatory function

Presently, how these protein-protein interactions relate to the tumor suppressor activity of merlin is not well understood. Merlin is located in the cell membrane and binds to transmembrane proteins and actin with its N-terminal FERM domain. Merlin’s ability to function as a growth regulator is also related to its ability to form such intramolecular associations [119]. Two such interactions have been identified. The first interaction is between residues that fold the N-terminal end of the protein onto itself, while the second interaction folds the entire protein so that there is no contact between N- and C-terminal ends of the protein [84, 105, 117]. In a fashion similar to the ERM proteins, merlin may cycle between the “open” and “closed” conformations that differentially determine whether it binds with the ERM proteins or other molecules to transduce merlin’s growth inhibition signal [120]. Merlin’s function is regulated by phosphorylation [108, 109]. Phosphorylation of the C-terminus disrupts N-terminal and C-terminal binding, which favors the “open” configuration of merlin. In this state, proliferation is favored. In the “closed” configuration, where the N-terminal FERM and C-terminal domains are bound to each other, merlin mediates contact-dependent growth inhibition.

The association of merlin with CD44 and B1-integrin raises the possibility that merlin might function as a molecular switch in the signaling pathways. CD44 is a transmembrane hyaluronic acid receptor implicated in cell-cell adhesion, cell-matrix adhesion, cell motility, and metastasis [92, 103]. Merlin mediates contact inhibition of cell growth through signals from the extracellular matrix. At high cell density, merlin becomes hypophosphorylated and inhibits cell growth in response to hyaluronate, a mucopolysaccharide that surrounds cells [121]. At low cell density, merlin is phosphorylated, growth permissive, and exists in a complex with ezrin, moesin, and CD44. CD44 is also thought to associate with anti-mitogenic signals, such as CD44-dependent MYPT1 phosphatase, which hypophosphorylate merlin at the C-terminus and maintain the closed configuration to inhibit cell growth. In this manner, CD44 may amplify the function of hyaluronate, which also maintains merlin in a hypophosphorylated state and signals growth arrest. These data indicate that merlin and CD44 form a molecular switch that specifies growth arrest or proliferation.

Various pro-mitogenic signals act via activating kinases that phosphorylate merlin, disrupt the binding between the N-terminal FERM and C-terminal domains, and allow for a growth-permissive configuration. For instance, Rac1, a member of the RhoGTPase family, has been demonstrated to promote the phosphorylation of merlin, thereby inactivating its growth suppressor mechanism [109, 110]. Among the effects of Rac1 and Cdc42 GTPases, members of the p21-activated kinase (PAK) family have demonstrated interactions with merlin. Specifically, Pak2 has been shown to phosphorylate merlin at serine 518 and inactivate its function [91, 105, 122, 123]. Other pro-mitogenic signals act by directly degrading merlin. The cyclic AMP-protein kinase A (PKA) pathway and AKT signaling pathways both modulate merlin phosphorylation and ubiquitination. Activating these pathways drives merlin into its open, growth-permissive configuration and tags merlin for proteasome-mediated degradation, both of which disrupt its tumor suppressor function [124].

Kissil et al. [121] also reported an interaction between merlin and Pak1, and merlin could inhibit the activation dynamic of Pak1. Loss of merlin expression leads to the inappropriate activation of Pak1, while overexpression of merlin results in the inhibition of Pak1 activity. Conversely, merlin is activated by dephosphorylation at serine 518, which occurs on serum withdrawal or on cell-cell or cell-matrix contact [119]. Although the members of the PAK family have been implicated in various cellular processes, such as cytoskeletal reorganization and apoptotic signaling, their exact roles and functions have not been clearly defined. Jin et al. [125] identified the myosin phosphatase (MYPT-1-PP1δ) as a merlin phosphatase. Interestingly, the cellular MYPT-1-PP1δ-specific inhibitor, CPI-17, could cause decreased merlin activity by merlin phosphorylation, Ras activation, and transformation. These results implicate MYPT-1-PP1δ and CPI-17 as important regulatory components in the merlin tumor suppressor pathway.

Merlin’s Growth Regulatory Function Is Related to Its Conformation and Protein-Protein Interactions

The activities of the ERM proteins are controlled by self-association of the proteins’ N-terminal and C-terminal regions [126, 127]. The ERM proteins can exist in the “closed” conformation, where the N- and C- terminal regions undergo an intramolecular interaction, thus folding the protein to mask the conserved actin-binding site (Fig. 10.3). The molecule can be converted to the “open” conformation in which the intramolecular interaction is disrupted by signals such as phosphorylation or treatment with phosphoinositides [107, 121, 128].

Fig. 10.3

The NF2 protein is divided into three general sections. The FERM domain is followed by an alpha-helical domain and a hydrophilic tail. Merlin can dimerize with itself and heterodimerize with other ERM family proteins. (Adapted from [79] and Adapted from Surgery of the Ear. 6th Ed. Eds. PMPH. 2010)

The association between merlin and HRS, a substrate implicated in the signaling pathway initiated by hepatocyte growth factor (HGF) binding to the c-met receptor [120], appears to be regulated by protein folding [105]. These results suggest that the ability of merlin to cycle between the “open” and “closed” conformations may integrate CD44 and HGF signaling pathways. All these findings are relevant to growth regulation. Also, merlin can exert its activity by inhibiting phosphatidylinositol 3-kinase (PI3-kinase) through binding to PIKE-L, a brain-specific GTPase that binds to PI3-kinase and stimulates its lipid kinase activity [109]. This finding suggests that PIKE-L is an important mediator of merlin growth suppression. Along this notion, we have found that the PI3-kinase/Akt pathway is activated in VS [129, 130].

The configuration of merlin also modulates the function of downstream proteins, such as B1-integrin. B1-integrin is involved in cell-cell adhesion and is a key protein that drives contact-dependent growth arrest. The N-terminal FERM domain of merlin interacts with the transcription factor Yes-associated protein 1 (YAP1) and with large tumor suppressor genes 1 and 2 (LATS1/2). LATS1/2 is thought to drive hypophosphorylation of YAP1, allowing for YAP1 relocation to the nucleus and inhibition of B1-integrin-dependent cell adhesion. When merlin is phosphorylated, by pro-mitotic kinases such as Pak2, FERM, and C-terminal domain interactions with YAP1 and LATS1/2 are reduced, resulting in YAP1 dephosphorylation and suppression of B1-integrin-mediated growth arrest. Thus, merlin phosphorylation is both mediated by upstream proteins and can regulate the function of downstream effectors involved in growth arrest and proliferation [131].

In addition, Lallemand et al. [108] showed that in Nf2^−/− mouse embryo fibroblasts, Nf2 deficiency led to piling up of cells, hyperproliferation, and defective cadherin-mediated cell-cell interactions. When functioning normally, these interactions terminate proliferation by contact inhibition through adherens junctions. Furthermore, merlin colocalizes and interacts with and causes maturation of adherens junction components in confluent wild-type cells. Stated simply, when normal Schwann cells contact each other, special signaling through these junctions stops further growth. In the absence of merlin, the adherens junctions are immature and do not stop the schwannoma cells from terminating their growth when they contact each other. These results indicate that merlin functions as a tumor suppressor at least in part by controlling cadherin-mediated cell-cell contact [132].

The NF2 Gene Promoter

Characterization of the NF2 regulatory regions is important for screening for mutations in both spontaneous and familial tumors in which no mutation was found in the NF2-coding region. We have shown that transcription of the NF2 gene is initiated at multiple start sites, and multiple regions in the NF2 promoter are required for full NF2 promoter activity [71, 133]. Both positive and negative regulatory elements required for transcription of the NF2 gene have been found in the 5′ flanking region of the promoter (Fig. 10.1). In particular, a sequence rich in guanine and cytosine nucleotide bases located in the proximal regulatory region, which can be bound by the Sp1 transcription factor, serves as a positive regulatory element [71]. This region is the area of the gene adjacent to the 5′ of the coding region which regulates, either positively or negatively, the translation of the DNA into messenger RNA as we have described. Both the 5′ and the 3′ flanking regions of the human NF2 locus are G/C rich and could serve as a target for gene methylation and inactivation as a form of therapy [70, 71].

We have also shown that a 2.4-kb human NF2 promoter could direct where the NF2 gene expression would occur as early as embryonic day 5.5 [88]. During early embryonic development, strong NF2 promoter activity was detected in the developing brain and in sites containing migrating cells including the neural tube closure and branchial arches. Interestingly a transient change of NF2 promoter activity during neural crest cell migration was noted. While little promoter activity was detected in premigratory neural crest cells at the dorsal ridge region of the neural fold, significant activity was seen in the neural crest cells already migrating away from the dorsal neural tube. In addition, we detected considerable NF2 promoter activity in various NF2-affected tissues such as the acoustic and trigeminal ganglia, spinal ganglia, optic chiasm, the ependymal cell-containing tela choroidea, and the pigmented epithelium of the retina. The NF2 promoter expression pattern during embryogenesis suggests a specific regulation of the NF2 gene during neural crest cell migration and further supports the role of merlin in cell adhesion, motility, and proliferation during development [88].

Alternatively Spliced NF2 mRNA Isoforms in Vestibular Schwannomas and Other Cell Types

The NF2 gene undergoes alternative splicing in the coding exons [42, 51, 71, 135,136,137,138,139,139]. Multiple alternatively spliced NF2 transcripts have been identified in various human cells. The most common isoforms in these cells were isoforms II (containing all 17 exons) and isoform I (without exon 16). Presently, the role of alternative splicing of NF2 mRNA is not well understood. It is possible that the functional contribution of the NF2 tumor suppressor may require a balanced expression of various isoform proteins in Schwann cells and/or other cell lineages [71, 140]. Alternative splicing may be another mechanism for Schwann cells to inactivate merlin function and/or to generate isoforms that have additional properties conducive to tumor formation.

Immunohistochemical Markers of Growth in Vestibular Schwannomas: Clinical Correlation

Attempts to correlate clinical parameters with immunohistologic evaluation of protein expression in VS have been performed. An increase in Ki-67, a protein, which is an index of nuclear proliferation, was shown to correlate with the growth of solid schwannomas on MRI [130, 131]. Higher rates of tumor recurrence have also been suggested in tumors with an increased rate of nuclear proliferation and mitotic indices, although the supporting data for this claim was not conclusive [141]. Positron emission tomography (PET) scanning has been conducted to assess the metabolic activity of VS preoperatively and to assess the metabolic activity within the proliferation index, Ki-67. However, no correlation was found. Additionally, there was no correlation between 18-fluorodeoxyglucose (FDG) uptake (indicator of metabolic activity in tissues) and Ki-67 expression measured by immunostaining [142]. This is most likely because VS are slow-growing tumors with less than 5% of tumor cells being in S-phase or active division [143].

Another possible marker for tumor growth is the transforming growth factor β1 (TGF-β1). Immunostaining for TGF-β1 was positive in 96% of blood vessels within schwannomas and in 84% of schwannoma tissue samples; however, no direct correlation with tumor growth was found [144]. Immunohistochemical association of β1-integrin with merlin has been demonstrated but has not been related directly to tumor phenotypes [97]. Cystic schwannomas are associated with a 36-fold decrease in nuclear proliferation as measured with Ki-67 staining when compared to solid tumors, suggesting that the rapid clinical growth seen in cystic schwannomas is related to the accumulation of cyst formation but not by an actual increase in the growth rate of tumor cells [145, 146]. Also, NF2-associated schwannomas have been shown to have an increased proliferation index by Ki-67 and proliferating cell nuclear antigen (PCNA) immunostaining, when compared to unilateral solid schwannomas [147, 148]. PCNA is a proliferation index marker and is important in cell division.

Recently, we detected higher levels of cyclin D₃ expression, which is associated with G₁ cell cycle progression, in five of ten VS, compared to Schwann cells in adjacent normal vestibular nerve [46]. In contrast, expression of the cyclin D₁ protein, which controls Schwann cell differentiation, was not detected in any of the schwannomas examined [149]. Another study of 180 patients with previously untreated VS, including sporadic cases, who underwent surgical resection found that cyclin D₃ was overexpressed in 44% of tumors and cyclin D₁ in 68% of tumors [150]. These results suggest a possible role for cyclin D₃, and potentially cyclin D₁, in the growth of some VS cells.

Taken together, these studies demonstrate relatively little correlation between clinical growth as assessed by MRI scans, historical data, and nuclear growth indexes in solid unilateral and NF2-associated schwannomas. Cyclic tumor growth appears to occur via a different mechanism. Although the defective NF2 gene is the underlying common denominator in tumor formation of all three tumor types, other differences at the molecular level likely account for the variable clinical presentations of these tumors.

Identifying Deregulated Genes in Vestibular Schwannomas

To further elucidate the growth pathways in VS, the gene expression profiles have been studied. The study of large-scale gene expression profiles utilizing the cDNA microarrays allows examination of the transcriptome of tissue and gives a means of exploring a broad view of the basic biology of tumors [151]. We studied gene regulation in VS using cDNA microarray analysis and found 42 genes, which were significantly upregulated including osteonectin, endoglin, and Rho B, when compared to normal vestibular nerve tissues [5, 152]. Additionally, multiple genes were found to be significantly downregulated in the majority of VS examined. Of these genes, a putative tumor suppressor gene LUCA-15 was downregulated in seven of eight schwannomas studied.

More recently, next-generation sequencing of sporadic VS has provided additional information about genetic alterations underlying tumorigenesis. In an analysis of fresh frozen tumor samples from 23 patients with sporadic VS, 31 unique variants in the NF2 gene were identified, of which 10 were splice site, 2 were missense, 7 were premature stop, and 11 were frameshift mutations. Several major chromosomal anomalies were also observed, including loss of a chromosome 15, 18p, 16q, or an X or Y chromosome, which were noted only in tumors with an aggressive phenotype [153]. A second study of 46 sporadic vestibular schwannomas found mutations in a number of genes, including 25 patients who expressed mutations in axonal guidance pathways, 5 patients with CDC27 mutations, 2 patients with USP8 mutations, and 35 patients with at least one NF2 mutation. Of the tumors analyzed, none showed microsatellite instability [154].

Osteonectin is a secreted glycoprotein that interacts with extracellular matrix proteins to decrease adhesion of cells from the matrix, thereby inducing a biological state conducive to cell migration. Endoglin, a TGF-β receptor-binding protein, was found to be significantly upregulated in all the solid tumors but not in any of the cystic tumors examined. It is likely that increased endoglin expression may induce downstream signaling proteins, somehow leading to an aggressive cystic phenotype [5]. An example of a deregulated signaling pathway suggested by the microarray data is the retinoblastoma protein (pRB)-cyclin-dependent kinase (CDK) pathway [155]. Among genes involved in G₁-S phase progression, CDK2 was found to be downregulated in seven of eight tumors. In addition, upregulation of transforming factor RhoB was found in all the schwannomas examined. Further examination of these deregulated genes as potential downstream targets of the NF2 tumor suppressor should provide us with targets for pharmacotherapeutic interventions.

Environmental Exposures

There is evidence that radiation may be associated with the induction of vestibular schwannoma growth. While other sites and tumors also show a propensity for growth in response to environmental exposure, the vestibular nerve appears to contain an increased sensitivity for tumor growth as compared to other regions within the central nervous system or elsewhere.

Evans and colleagues note an 18.8-fold increased relative risk of schwannoma induction and a 9.5-fold increased relative rate of meningioma induction following radiation doses of 2.5 Gy from the treatment of tinea capitis, adeno-tonsillar disease, and capillary hemangiomas [156]. The radiobiologic effect was dose dependent similar to the tissue effect and tumor growth response observed in post-WWII Japan. Preston and colleagues [157] showed that VS were the most abundant intracranial tumors encountered, with a high propensity for Schwann cell mutation in response to radiation damage.

Overall, the risk of tumors induced by therapeutic radiation appears to be approximately 0.1–3% after 30 years [146, 158]. In light of the 33–40% lifetime risk of malignancy in the general population, Evans [146] argues that the benefits of single-dose radiation treatment for VS may be justified for those with documented tumor growth that refuse surgery or for the elderly or infirm. However, patients with NF2 are disproportionate recipients of radiation-induced malignant tumors. Although only 7% of patients with VS have NF2, 50% of malignant changes induced by stereotactic irradiation of VS occur in NF2 patients. We strongly caution against the use of radiation treatment for benign tumors in childhood and in tumor-prone conditions such as neurofibromatosis.

Cellular Telephone Risk

Several studies addressed the risk of cellular telephone use and the formation of VS presumably secondary to electromagnetic radiation over a long exposure period. Hardell et al. [159] reviewed 13 case-controlled studies and found 9 case-control studies to have reported patients with over 10 years of cellular telephone use. They concluded that the odds ratio for ipsilateral cellphone use and the development of VS was 2.4 (95% confidence interval 1.1–5.3). Additional case-control studies suggest increased risk for vestibular schwannoma and glioma with risk elevation greatest for high-grade gliomas associated with unilateral long-term exposure [155, 160]. However, other studies have found no difference in VS size or incidence with long-term exposure and instead suggest that cell phone use may be correlated with earlier diagnosis of VS [161, 162].

Genetic Screening for NF2

A patient with NF2 has a 50% chance of passing the affected NF2 allele along to their offspring if (1) their spouse does not have NF2 and if (2) they do not have a mosaic form of NF2. (Mosaicism occurs when a mutation in the NF2 gene occurs after an embryo has undergone several sets of cell division. Therefore, only a portion of the child’s cells carry a mutation in the NF2 gene.) A child who inherits an abnormal copy of the NF2 tumor suppressor gene has a 95% chance of developing bilateral VS.

Because early intervention is very important for clinical outcomes, genetic and clinical screening for at-risk patients is strongly advocated for those at risk to develop NF2. This includes anyone with a first-degree relative with NF2, patients under 30 years old with a unilateral VS, or any patient with multiple intracranial, spinal, or peripheral skin tumors that are associated with NF2. Any offspring of patients with NF2 should have annual ophthalmology examinations starting soon after birth since cataracts can begin at a very early age. Annual neurology examinations should be started at 7 years of age or earlier if neurologic deficits are noted. We recommend biannual audiograms and annual MRI evaluations beginning at age 7. Others have recommended starting a similar screening process at 10 years of age with an MRI every other year and annually once a tumor is found [163].

Mutation screening for the first-degree relatives of NF2 patients is recommended when the probands’ mutation has been identified either from a schwannoma or from leukocyte screening [157, 164]. Other family members at risk can be identified and screened for developing NF2 or receive appropriate genetic counseling. Patients who are first-degree relatives of NF2 patients where the mutation in the NF2 gene has not been identified may have more ambiguity regarding the screening of their NF2 gene for mutations if no mutation is localized. Appropriate audiometric testing and neural imaging methods need to be carried out as false negatives in mutation screening can occur in up to 25% of patients. Annual MRI screening and biannual audiometric testing would still be recommended for children older than 7 years of age who had a negative test. Additionally, if a mutation was detected during DNA sequencing, we recommend annual MRIs and biannual audiometric testing in order to detect the development of VS at the earliest possible stage. Early detection in NF2 does make a significant difference in the ability to successfully preserve hearing and facial nerve function.

When an NF2 mutation has already been identified in an affected family member, screening additional first-degree family members for the same mutation is considerably less costly and is recommended. The sensitivity of genetic testing in this circumstance is extremely high since the DNA being screened can be directly compared to the known mutation in the affected family member. In this case, knowing with near certainty that a child does not carry an NF2 mutation can avoid the frequent MRI examinations that would otherwise be required [48, 158]. Therefore, if a family has multiple members diagnosed with NF2 and mutation testing has not been performed, it is worthwhile to identify the mutation in at least one family member so that future family members can be screened. This is most effectively done when tumor tissues are available for mutation screening. However, biopsy for genetic diagnosis alone is not suggested if tumor removal is otherwise not indicated as clinical screening protocols are also effective.

Potential Drug Treatments

As the understanding of merlin and its interacting partners is better, new means of targeting these pathways for intervention in schwannoma cells is occurring. Some treatments are aimed at restoring the wild-type merlin protein to the mutant cells, while other strategies attempt to block the proliferation of schwannoma cells by blocking its growth pathways or by increasing programmed cell death or apoptosis. For example, Messerli et al. [165] showed an effective reduction of schwannoma growth in both a transgenic and a xenograft mouse model by injecting wild-type merlin packaged in an oncolytic recombinant herpes simplex virus vector in the tumor. The vehicle for delivery was a replication competent herpes simplex virus, which has tumoricidal properties. There were no apparent toxicities with the injection of the oncolytic vector. A paradigm for the use of oncolytic vectors to reduce the volume of benign tumors when surgical resection may cause nerve damage is suggested. Likewise, reintroducing the NF2 gene by direct injection resulted in tumor regression and reversion toward normal histology [166].

Hansen et al. [167] and Doherty et al. [168] demonstrated cell surface human epidermal growth factor, also known as erythroblastic leukemia viral oncogene homolog 2 or ErbB2, receptors as possible targets for VS cell growth inhibition. Blocking the ligand or interacting proteins, which activates this receptor by treatment with blocking antineuregulin antibodies, also inhibited schwannoma cell proliferation. ErbB2 is a tyrosine kinase receptor, a common cellular switch, on the VS cell surface, which triggers the MAPK kinase cascade, a process that leads sequentially to nuclear activation of cell division. One of its ligands, neuregulin-1, is a growth factor produced by schwannoma cells, thus creating an autocrine feedback loop; the schwannoma cells cause further stimulation of schwannoma growth.

Plotkin et al. showed that epidermal growth factor receptor (EGFR) blockage with erlotinib HCl was tolerated and showed radiographic and audiometric response to treatment in a patient with NF2 in short follow-up. Karajannis et al. demonstrated effectiveness of lapatinib, an Erb2 inhibitor for control of the growth of the majority of NF2-associated vestibular schwannomas although only 4 of 17 showed reduction in volume by greater than 15% [170].

Several investigators demonstrated that intervention at the intracellular proliferation pathways, specifically the MAPK pathway and the PI3K pathway, respectively, may also result in schwannoma growth suppression. Nakai et al. [169] showed that merlin is involved in suppression of Rac signaling, and cultured schwannoma cells contain elevated, GTP-bound, active Rac. Application of a Rac-specific inhibitor, the chemical compound NSC23766, to schwannoma cells restored neuronal interaction. The data support the significance of regulated Rac signaling in mediating Schwann cell-axon interaction and suggest controlling Rac activity as a possible therapy for schwannomas.

Jacob et al. [170] demonstrated the activation of the PI3K/AKT pathway in human schwannomas. It has been shown that merlin can inhibit PI3K through its binding to the PI3-kinase enhancer-L (PIKE-L), a GTPase which binds to and stimulates PI3K (Fig. 10.4). The suppression of PI3K activity results from merlin disrupting the binding of PIKE-L to PI3K. Our recent studies using cultured schwannoma cells further suggest that both an AKT pathway inhibitor and a histone deacetylase inhibitor could suppress schwannoma cell growth by blocking the PI3K pathway. Utilizing a quantifiable xenograft model for VS, we are currently testing the efficacy of these novel compounds. Recently, Zhao et al. performed a screening of 157 inhibitors targeting the PI3K/AKT pathway in an in vitro NF2 model and identified over a dozen inhibitors which suppressed schwannoma growth. Of these inhibitors, AZD8005 and PQR309, which are orally available, were found to inhibit growth both in vitro and in vivo. These findings highlight potential therapeutic benefit to targeting proteins within the PI3K/AKT pathway [171].

Fig. 10.4

PI3K/AKT signaling. Insulin and other growth factors activate cell surface tyrosine kinase receptors that stimulate PI3K activity. PI3K converts PIP2 to PIP3. PTEN is a negative regulator of the AKT pathway that catalyzes the conversion of PIP3 back to PIP2. PDK1 and PDK2 maximally activate AKT. Phosphorylation of downstream effectors of the AKT pathway inhibits these proteins. Existing as a complex, TSC1 and TSC2 normally inhibit mTOR activity. However, when TSC2 is phosphorylated by AKT, it dissociates from TSC1, and mTOR becomes active. Active GSK and FOXO function within the nucleus as transcriptional regulators that activate genes promoting apoptosis. Merlin, the NF2 gene product, can bind PIKE-L and inactivate PI3K. A small-molecule PDK1 inhibitor can inhibit phosphorylation of AKT at the 308 site. (PNAS 2004). (Adapted from Surgery of the Ear. 6th Ed. Eds. PMPH. 2010)

Transduction of growth factor stimulation of the mammalian target of rapamycin (mTOR) signaling pathway ultimately results in stimulation of protein synthesis and entry into the G₁ phase of the cell cycle. Decreased phosphatase and tensin homolog gene (PTEN), a human gene that acts as a tumor suppressor gene, manifests as AKT and mTOR activation stimulating growth. Additionally, hypoxia-inducible factors HIF-1 alpha and HIF-2 alpha show increased expression and account for enhanced angiogenesis with mTOR activation. This may lead to the necessary blood vessel supply to schwannomas. AKT and mTOR activation have been noted in VS. The immunosuppressant, rapamycin, inhibits mTOR and induces G₁ arrest and apoptosis. Newer mTOR inhibitors, such as temsirolimus, have shown good effect in the treatment of renal cell carcinoma alone and in combination with other agents (Fig. 10.5). Everolimus (or Rad001) is an orally administered mTORC1 inhibitor. Goutagny et al. in a prospective trial of everolimus for progressive VS in NF2 patients showed five of ten patients with decrease in growth rates and resumption of growth after termination of drug [172].

Fig. 10.5

Alternative pathways. In this schematic in a recent review by Dan Scoles, merlin’s interaction with the ErbB2-paxillin complex, PIKE, and elF3c, its effects on cyclin D1 expression, and functional consequences of merlin loss suggest that signaling in the ErbB2-PI3K-mTOR-elF3c-cyclin D1 pathway may contribute to NF2 pathogenesis. PIKE-L enhances PI3K, which leads to phosphorylation and activation of PDK1 and PDK2, which in turn activates mitotic machinery. Drugs targeting this pathway therefore may be useful for inhibition of NF2 tumors with overactive mTOR signaling, including trastuzumab (Herceptin, LY294002, AKT inhibitors, UCN-01, Rad-001, CCI-779, and AP23573), small-molecule AKT pathway, and histone deacetylase inhibitors, and HSP90 inhibitor 17-allylamino-17-demethoxygeldanamycin (17AAG) may affect Akt and C-Raf, as well as Cdk4 and Cyclin D-1 (NexGenix). (Adapted from Surgery of the Ear. 6th Ed. Eds. PMPH. 2010)

Karajannis et al. reported the results of a separate trial using everolimus in ten patients with NF2 and progressive vestibular schwannomas. No objective improvement in hearing or imaging response was reported [173]. Subsequently, in a phase 0 presurgical clinical trial, blood levels of everolimus were found to be in a high therapeutic range. Tumor tissue drug concentration was determined. Only partial inhibition of phospho-S6 was found in the treated tumors, indicating incomplete target inhibition compared with control tissues from untreated patients. Everolimus led to incomplete inhibition of mTORC1 and downstream signaling. The authors concluded that their data may explain the “limited antitumor effect of everolimus observed in clinical studies for patients with NF2” and that such preclinical surgical studies may influence future preclinical and clinical studies [174]. Side effects of everolimus include mouth ulcers early in the treatment course, anemia, decreased neutrophil and lymphocyte counts, peripheral edema, hyperglycemia, and pericardial effusion [175].

Angiogenesis inhibitors such as bevacizumab (Avastin) have been used alone or in combination for treatment of colorectal, lung, and prostate carcinomas [177,178,179,179]. It acts by inhibiting vascular endothelial growth factor (VEGF) and is used in several combination treatment protocols. Bevacizumab has shown temporary suppression of VS growth. Its temporary effect on tumor growth reduction, the potential side effects of leukopenia, and bleeding limit its use in surgical patients, and it is currently considered a temporizing agent.

Plotkin et al. reported a cohort of ten NF2 patients with poor candidacy for surgery or radiation therapy who received intravenous bevacizumab over 18 months. A volume reduction of 20% or more was observed in 41% of tumors, primarily within the first 12 months of treatment, with relapse and regrowth following discontinuation of bevacizumab treatment [180]. Significantly favorable effects were observed in patients above 25 years of age, while younger patients with faster-growing tumors showed a weaker response to bevacizumab. A second retrospective analysis of 31 VS patients receiving bevacizumab found a decrease in tumor volume of 20% or more in 55% of the sample and an increase in word recognition score in 57% of the sample. At 1 year, 88% of patients had stable or decreased tumor size [5]. Reduction in tumor volume and improvement in hearing response with bevacizumab have been reported in several other cohort studies, demonstrating that bevacizumab may have a significant therapeutic effect in VS in about one-third of patients [182,183,183]. Adverse effects including hypertension, proteinuria, hemorrhage, and tumor growth with interruption or discontinuation of therapy are all known side effects to be managed [184, 185].

Advancements in Nonpharmacological Methods

Gene delivery methods have shown promise in the treatment of various genetic conditions, such as heritable hearing loss and spinal muscular atrophy. Such methods may provide benefits in NF2 cases with clear molecular targets. One approach involves the direct intratumoral injection of viral vectors carrying tumor-lytic agents such as caspases. In a mouse model with subcutaneously implanted immortalized human NF2 schwannoma cells, intratumoral injection of a herpes simplex virus-1 (HSV-1) vector encoding caspase-1 (ICE) under a Schwann cell-specific promoter (P0) led to reduction in tumor size [186]. A follow-up study demonstrated tumor regression with intratumoral injection of an adeno-associated viral (AAV) vector expressing ICE under P0 in schwannoma cells implanted into the sciatic nerve of mice [187]. Additional studies have demonstrated efficacy with AAV vector delivery of other molecular targets, such as speck-like protein, a pro-apoptotic schwannoma tumor suppressor, which led to reduced tumor growth and improved tumor-associated pain in a human xenograft model of NF2 [188]. Gene delivery may also involve nonviral vector methods, such as liposomal or peptide-based delivery systems. Liposomal delivery methods have been used in mouse models of genetic deafness. Delivery methods are evolving and may benefit from advancements in surgical technique, such as injection approach through the round window membrane, and studies to assess the safety and efficacy of viral and nonviral gene delivery methods.

Recent clinical studies have explored the use of immunotherapy methods, such as selective activation of cytotoxic T lymphocytes (CTLs), to target NF2 schwannomas. In a cohort of seven patients with NF2, Tamura et al. demonstrated that induction of VEGFR-specific CTLs via vaccination was tolerated without severe adverse effects. Two patients showed increases in word recognition scores, and tumor size reduction greater than 20% was observed in two patients, suggesting the preliminary efficacy of VEGFR-specific vaccination [189].

The medical treatment of slow-growing benign tumors must maintain efficacy over decades of potential treatment. Perhaps combined modality treatments might be considered. Two agents we are currently studying (a histone deacetylase inhibitor and an AKT pathway inhibitor) have shown the capacity to modulate radiation response in an additive fashion. By sensitizing targeted tissues, radiation doses could be lowered to optimistically promote a killing effect of tumor cells, reduce adverse effects on surrounding structures, preserve physiologic function, and decrease the risk of tumor promotion or malignant transformation.

Schwannomatosis (OMIM #162091), a distinct form of neurofibromatosis, is characterized by multiple schwannomas, but this condition most often lacks any VS. Patients with schwannomatosis frequently present with intractable pain rather than cranial nerve deficits. They seldom develop other intracranial tumors or malignancies. MacCollin et al. [190] noted that about one-third of patients with schwannomatosis had tumors in an anatomically limited distribution, such as a single limb, several contiguous segments of the spine, or one-half of the body. Sporadic cases of schwannomatosis are as common as NF2, but few cases of familial schwannomatosis have been identified in contradistinction to NF1 and NF2, which are autosomal dominant and highly penetrant. The underlying molecular disruption in schwannomatosis is a pattern of somatic NF2 gene inactivation incompatible with NF1 or NF2, associated with alterations in the SMARCB1 or LZTR1 tumor suppressor genes in familial disease [191, 192].

Genetics of Paragangliomas

Paragangliomas of the head and neck are rare, mostly benign tumors. The hereditary paraganglioma syndromes are autosomal-dominant conditions with increased risk for multifocal tumors of the sympathetic and parasympathetic neuroendocrine systems. Approximately 30% of paraganglioma patients harbor a germ line mutation, which is carried by all the cells of the body (as opposed to a somatic mutation, which is only found in the tumor tissues). In more than a third of these patients, there is no prior family history, thus representing newer mutations.

About 10–15% of paragangliomas are caused by mutations in the succinate dehydrogenase (SDH) gene or its anchoring subunits B (SDHB), C (SDHC), or D (SDHD) [193]. They may be hormone secreting. Malignant paragangliomas have been partially associated with mutations in SDHB but have also been reported in SDHD [194]. The SDH genes provide a protein that is a small part of the cytochrome b complex in the mitochondrial complex H, one of the five protein assemblies or complexes that form the electron transport chain. By production of reactive oxygen molecules, this pathway is considered important in the signaling and detection of oxygen tension. The gene encoding SDHD consists of 4 exons and yields 159 amino acid polypeptides [195].

Other mutations found to be associated with paragangliomas include the prolyl hydroxylase domain (PDH) proteins, which play a major role in regulating HIF, which induces expression of genes involved in angiogenesis, erythropoiesis, and cell metabolism, proliferation, and survival. Mutations in the prolyl hydroxylase domain 2 gene (PHD2) have recently been described in a patient with erythrocytosis and recurrent paragangliomas [196].

Summary

Our understanding of the molecular mechanisms underlying VS formation is advancing. Elucidating merlin’s function and its interactions with other proteins and signaling pathways and regulation of the NF2 gene will accelerate the development of novel treatments for VS, such as drug therapies designed to stop schwannoma progression. This may offer alternatives to the current options of untreated observation of tumor growth, stereotactic radiation, or surgical removal.