Pathogenic Variants and Allele Loss of the NF2 and LZTR1 Gene in Sporadic Vestibular Schwannoma

Abstract

Background/Aim

Pathogenic variants and allele-loss of the NF2 gene with Merlin loss as consequence is the driving genetic event for vestibular schwannoma development. Our knowledge about the pathogenic NF2 variants in sporadic vestibular schwannoma is insufficient. Therefore, we analyzed a cohort of sporadic vestibular schwannomas by panel-sequencing.

Patients and Methods

Forty-one sporadic vestibular schwannomas from 26 male and 15 female patients were included. DNA from tumor tissues was sequenced with a custom panel for the NF2 and LZTR1 genes. Allele-loss of the NF2 locus was also examined using multiplex-ligation-dependent probe-amplification. These genetic data were correlated with clinical parameters including hearing, tumor extension and growth.

Results

Among the 41 tumor samples, 34 had one pathogenic variant or an allele-loss of NF2 gene and one tumor showed a pathogenic variant in the LZTR1 gene. Allele frequencies of the total of 46 pathogenic variants varied from 0.05 to 0.82, and none of these variants was found in blood. For 6 tumors, no pathogenic variants were found while 4 of them had allele-loss of the NF2 gene. When the tumors were divided into 3 groups according to the counts of inactivating events (pathogenic variants and allele loss), the clinical parameters including hearing, tumor structure in MRI, tumor growth, tumor size and postoperative facial function did not differ significantly.

Conclusion

There was no correlation between phenotype and genetic alterations of the NF2 or LZTR1 gene in sporadic schwannomas. Genetic inactivating events are the precondition for the development of vestibular schwannomas but do not influence their growth and other features.

Introduction

Vestibular schwannoma are benign tumors of the vestibular nerve in the cerebellopontine angle. Most of them arise sporadically and only about 5% grow in patients with neurofibromatosis type 2 (NF2), a genetic disorder with an inherited inactivating variant of the NF2 gene, which codes for the tumor suppressor Merlin. Merlin’s loss of function is the main known pathogenic factor in vestibular schwannoma pathogenesis. Our knowledge regarding genetic alterations in sporadic vestibular schwannoma is still insufficient, though Merlin inactivation is known as the main pathogenic factor in these tumors (1).

In patients of neurofibromatosis type 2 (NF2), mosaicism is frequent which means that the de novo pathogenic NF2 variants are only in subgroups of the cells, depending on the time point in embryonic stage when the variant occurred. The hypothesis is, that sporadic vestibular schwannoma represents an extreme form of mosaic NF2 where the de novo pathogenic variants occur extremely late and, therefore, are only carried in an extremely small number of cells.

Target sequencing using custom panels enables effective detection of variants for hundreds of samples. Furthermore, it is far more sensitive than the conventional Sanger sequencing and can detect variants in low allele-frequencies. This is important for tumor tissues which may contain a large amount of non-tumor cells. In this study, we applied a custom panel covering the entire coding region of the NF2 and leucine zipper-like transcriptional regulator 1 (LZTR1) genes for 41 sporadic vestibular schwannomas. In addition, we applied the multiplex ligation dependent probe amplification (MLPA) to assess copy number variations of all exons of NF2 and two exons of the LZTR1 gene.

Patients and Methods

Tissue samples and clinical data. The study was conducted according to the guidelines of the Declaration of Helsinki and approved on 17/03/2021 by the Institutional Review Board of the University Hospital Wuerzburg (#241/20). Written informed consent was obtained from all patients for the use of their tissue in this study. All patients were treated in the Neurosurgery Department of the University Hospital Wuerzburg between 2021 and 2022. Directly after surgical excision, the tissue was processed for DNA extraction. All samples were neuropathologically assessed according to EANO guidelines and WHO criteria (1,2). Forty-one tumors were diagnosed as sporadic vestibular schwannoma and among them 6 were recurrences. For 32 out of the 41 patients, blood DNA was available and also analyzed by panel-sequencing as described below.

Patient clinical data was collected retrospectively (Table I, Table II). Hearing function and tumor extensions were categorized using the Hannover classification (3,4) and tumor growth dynamics were classified by magnetic resonance imaging during a “watch and wait” period before surgery if available (3,5). Tumor growth of more than 2 mm in a year was categorized as rapid growing and less as slow growing. Vestibular schwannomas with a homogenous contrast enhancement were classified as homogenous, tumors with cystic components were categorized as cystic and with irregular contrast enhancement as inhomogeneous. Radiosurgery or bevacizumab treatment before surgery was defined as pretreatment.

Table I. Summary of patients’ clinical parameters.

Table II. Phenotype of the patients including symptoms, function, tumor development, tumor extension and pre-treatment.

Sex: 1: female, 2: male; side: 1: tumor on the right side; 2: tumor on the left side; Tumor extension according Hannover classification: 1: T1; 2: T2; 3: T3a; 4: T3b; 5: T4a; 6: T4b; Rapid vs. slow growing: 1: rapid; 2: slow; 3: no follow up available; Antoni Type: 1: Type A; 2: Type B; 3: Type A/B; Pretreatment regarding the vestibular schwannoma: 1: yes; 2: no; Hearing according to Hannover Classification: 1: H1; 2: H2; 3: H3; 4: H4; 5: H5; 6: H6; Facial function according to House Brackmann: I: HB1; II: HB2; III: HB3; IV: HB4; V: HB5.

DNA extraction and panel-sequencing. Total DNA was extracted from native tissue and from ethylendiaminte-traacetat (EDTA)-blood from the patients utilizing the Gene Matrix Universal DNA Purification Kit (Roboklon, Berlin, Germany). Purified DNA samples were stored at –80˚C and subjected to targeted sequencing for the NF2 and LZTR1 genes using a custom panel of amplicons covering the entire coding and splicing sequences of the two genes. These amplicons were prepared into a library using an Illumina Ampliseq Plus kit (Illumina, Berlin, Germany). The libraries were sequenced on an Illumina iSeq100. The resulting reads were evaluated by an integrated “amplicon analysis module” (Illumina iSeq 100, v2.1.0) and the variants that deviated from the reference sequence were specified and further evaluated manually.

Multiplex ligation-dependent probe amplification (MLPA). In order to examine the copy number variations of exons and the entire gene an MLPA (multiplex ligation-dependent probe amplification) analysis was carried out for the NF2 genes according Kluwe et al. (12). The data were evaluated using the analysis software Coffalyser (MRC-Holland, Version 240129.1959).

Statistical analysis. All statistical computations were performed with Graph Pad Version 9 (GraphPad Software, San Diego, CA, USA). Normality was tested by the Shapiro-Wilk test. Statistical significance was determined using the Mann-Whitney-U-test and the Kruskal Wallis test. p<0.05 was considered as statistically significant. Correlation was evaluated using the Pearson correlation coefficient.

Results

Basic clinical and genetic features of the tumors. A total of 41 vestibular schwannomas from 26 female and 15 male patients (mean age 50.5±13 years) were assessed for pathogenic variants of the NF2 and LZTR1 genes. All tumors were sporadic ones, meaning that none of the patients met the diagnostic criteria for NF2-related schwannomatosis. Among the 41 tumors, 35 were primary and 6 were recurrences. The clinical parameters of the patients are summarized in Table I. A total of 46 variants were found and none of them was detected in the available blood samples of 32 of the patients, ensuring that all these pathogenic variants are somatic. The allele frequencies of the somatic variants varied from 0.05 to 0.82 while 35 variants had allele frequencies below 50%. Especially, 12 (26%) variants had allele frequencies below 20%.-All pathogenic variants and allele loss of the NF2 and LZTR1 genes are summarized in Table III.

Table III. Type and depth of each pathogenic variant and its effect.

Clinical features versus counts of inactivating events in each tumor. Among the 41 tumors, 12 grew fast, 18 had a cystic or inhomogeneous appearance in magnet resonance imaging (MRI), 15 showed worse hearing or functional deafness and five patients had a pretreatment with a surgery before (Table II). One of these patients had, years before surgery, a chemotherapy and stem cell transplantation because of blood cancer as a child. In further analysis, the 41 tumors were divided into three groups, according to the counts of inactivating events in each of them (see Table IV): Group 0=tumors with no inactivating event (n=4); Group 1=tumors, each with one inactivating event (pathogenic variants or allele loss), including the tumor with unknown status of allele loss (n=8); Group 2=tumors with two or more inactivating events each (n=29). Regarding hearing (Figure 1B), a significant difference was found between Group 0, Group 1 and Group 2 with score 1.5 versus 2.25 and 3.5, respectively, and p=0.015.

Table IV. Combination of genetic events in the 41 tumors analyzed.

Figure 1.

Phenotype correlation with the number of pathogenetic variants (PV). A) Facial function according to House & Brackmann (HB) Grade; B) Auditory function according Hannover Classification; the difference between 1 PV and 2 PV was statistically significant (p=0.015). C) Tumor growth dynamic (1: rapid; 2: slow); D) Tumor extension according Hannover Classification.

Tumor size (Figure 1D) did not correlate with the counts of inactivating events. The mean size score was 4.2 (T3B-T4A) in Group 0, 4.2 (T3B-T4A) in Group 1 and 4.96 (T4A) in Group 2. These differences were not significant according to Kruskal Wallis test (p=0.18). When slow growing was coded as 2 and fast growing as 1, Group 0 had a score of 1.7 and was not significantly slower than the score of 1.6 of Group 1 or the score of 1.3 of Group 2 (Figure 1C). There were no differences in the preoperative facial function. By contrast, worse postoperative results were found more frequently in Groups 1 and 2, compared to Group 0, although the difference was not significant (p=0.37) (Figure 1A).

One tumor of Group 1 and three tumors of Group 2 were treated with radiation or medical before the surgery. Thus, in our small cohort the number of pathogenic variants increases with pretreatment. There were no differences regarding resection extension or postoperative regrowth. For all four tumors in group 0, MRI showed completely homogeneous signals. In contrast, inhomogeneous or cystic signals were found in 1/8 (13%) of group 1 and in 9/29 (31%) of group 2 tumors. Thus, the proportion of inhomogeneous tumors seems to increase with the counts of pathogenic variants.

Within group 2 tumors, some had two pathogenic variants, and some had one pathogenic variant plus an allele loss. However, clinical parameters including hearing, tumor structure in MRI, tumor growth and postoperative facial function were similar. Tumor size showed differences and there was a trend, but it did not reach significance (p=0.06).

Unexpected genetic findings and clinical presentation. In tumor 18 a frame-shifting variant was found in exon 20 of the LZTR1 gene with an allele frequency was 66% (depth 508) and no pathogenic variant was found in the NF2 gene. MLPA detected an allele-loss of the entire NF2 locus. The two probes for two exons LZTR1 also showed copy number reduction, indicating that the allele loss covers the LZTR1 locus.

So, tumor #18 had a pathogenic frameshift variant of the LZTR1 gene. The tumor had medium size, and the preoperative hearing function was also medium. MRI signals were homogeneous. The resection was complete and there has been no recurrence so far. Histology revealed a typical schwannoma.

Another unexpected finding was in tumor 41. Two pathogenic variants plus an allele loss of the NF2 gene were found. The two frameshifting variants are in exons 12 and 15 with allele frequencies of 44% and 8%, respectively. The allele loss appeared heterozygotic in the MLPA data (Figure 2). So, tumor #41 had a high number of pathogenic variations. Although it was a large tumor it resulted in a medium hearing function preoperative. The tumor was inhomogeneous in MRI, the patient underwent surgery within three months after diagnosis. He received no pre-treatment and had no other tumors. Immediately after the operation, he had a facial palsy House Brackmann (HB)˚III, but so far (after 1.5 years of follow up) no recurrence was observed.

Figure 2.

MLPA of tumor #41 reveals an allele loss of the NF2 gene, which appeared heterozygotic.

Discussion

Using a panel covering the NF2 and LZTR1 gens, a total of 46 pathogenic variants were found in the 41 sporadic vestibular schwannomas. Panel sequencing has much higher sensitivity than the conventional Sanger sequencing and, therefore, can detect more variants in DNA from tumors, which often have non-tumor tissues of high proportion (6,7). Indeed, 12 variants found in this study had allele frequencies below 0.2, which would likely not have been detected by Sanger sequencing. According to the standard protocol for detecting somatic variants, only those in more than 5% of the total reads were recorded in the present study. However, it is still well possible that some specimens had extremely high proportion of non-tumor tissues and therefore pathogenic variants were in even lower allele frequencies which were not recorded. This may explain a part of missing pathogenic variants.

Missing pathogenic variants can also be explained by deep intronic variants, which are not covered by the applied panel. Deep intronic variants may cause alteration in the splicing and, therefore, can be pathogenic. A third explanation for not finding an inactivating event is, that the investigated piece of the tumor, contained a large portion of tumor-free tissue. This is due to the limited sensitivity of MLPA for detecting copy number variation. A fourth possible explanation of not finding an inactivating event is that the events may occur in other genes. If our panel did not include the LZTR1 gene, we would miss this pathogenic variant. A recent genome-wide association study found that the 9p21.3 region is associated with risk of vestibular schwannomas (8). Thus, inactivating events may be in genes in those regions.

Though LZTR1-related schwannomatosis is more closely associated with non-cerebral tumors, vestibular schwannomas can also develop. Pathogenic LZTR1-variants have even been found in the blood as germline variants in 4 (3%) out of 161 patients with sporadic vestibular schwannomas (9). Therefore, our finding of a pathogenic LZTR1-variant in a sporadic vestibular schwannoma is in concordance with the previous findings.

There is no straightforward correlation between the type nor the counts of the inactivating events and clinical features in this cohort of sporadic vestibular schwannomas. Correlation of genotype with disease-severity in NF2-related schwannomatosis (10-12) does not apply to sporadic vestibular schwannomas. The type of genetic variant influences the patient age at which a tumor develops and the number of tumors a patient develops. However, growth and associated clinical features are more biological and may not be influenced by genetic but rather non-genetic factors.

Carlson et al. (13) found that only major chromosomal abnormalities correlated with an aggressive phenotype. Havik et al. (14) investigated the genetic landscape of sporadic vestibular schwannoma with whole genome sequencing and found a pathogenic NF2-variants in 35 out of 46 cases. In 16 cases, they found two mutational hits, but nothing is reported about the chronological sequence of the mutations and a higher number of pathogenic variants could not be linked to radiosurgery. Sporadic spinal schwannomas also show pathogenic variants in the NF2 gene and two mutations in one tumor have been reported by Carvalho et al. (15).

Conclusion

There was no correlation between clinical phenotype and genetic inactivating alterations of the NF2 gene in sporadic schwannomas. Inactivation of the LZTR1 gene is also involved in the development of sporadic vestibular schwannoma. Genetic inactivating events are the precondition for the development of vestibular schwannomas but do not influence their growth and other features.

Funding

The University of Würzburg supported this work in the funding program Open Access Publishing.

Conflicts of Interest

There are no conflicts of interest.

Authors’ Contributions

All Authors contributed to the study conception and design. The study was supervised by MB and LK. Tumor tissue samples were provided by RE, CM, ML and CMM. Experiments were performed by LK, and data analyzed by TS, MB and LK. TS, MB and LK wrote the draft of the manuscript, which was subsequently revised by CMM, CM, ML and RE. All Authors read and approved the final manuscript. MB is the corresponding author.