Mutation profiling of anaplastic ependymoma grade III by Ion Proton next generation DNA sequencing

Ependymomas are glial tumors derived from differentiated Background: ependymal cells. In contrast to other types of brain tumors, histological grading is not a good prognostic marker for these tumors. In order to determine genomic changes in an anaplastic ependymoma, we analyzed its mutation patterns by next generation sequencing (NGS). Tumor DNA was sequenced using an Ion PI v3 chip on Ion Methods: Proton instrument and the data were analyzed by Ion Reporter 5.6. NGS analysis identified 19 variants, of which four were previously Results: reported missense variants; c.395G>A in , c.1173A>G in , IDH1 PIK3CA c.1416A>T in and c.215C>G in . The frequencies of the three KDR TP53 missense mutations ( c.1173A>G, c.1416A>T, , PIK3CA KDR TP53 c.215C>G) were high, suggesting that these are germline variants, whereas the variant frequency was low (4.81%). However, based on its IDH1 FATHMM score of 0.94, only the variant is pathogenic; other variants IDH1 , and had FATHMM scores of 0.22, 0.56 and 0.07, TP53 PIK3CA KDR respectively. Eight synonymous mutations were found in , , FGFR3 PDGFRA , , , , and genes. The mutation in EGFR RET HRAS FLT3 APC SMAD4 FLT3 p.(Val592Val) was the only novel variant found. Additionally, two known intronic variants in were found and intronic variants were also found in KDR and . A known splice site mutation at an acceptor site in ERBB4 PIK3CA , a 3’-UTR variant in the gene and a 5’_UTR variant in the FLT3 CSF1R gene were also identified. The p-values were below 0.00001 for SMARCB1 all variants and the average coverage for all variants was around 2000x. In this grade III ependymoma, one novel synonymous Conclusions: mutation and one deleterious missense mutation is reported. Many of the variants reported here have not been detected in ependymal tumors by NGS analysis previously and we therefore report these variants in brain 1 2 3 4


Introduction
Ependymal cells are macroglial cells which line the ventricles, the central canal of the spinal cord and form the bloodcerebrospinal fluid barrier, being involved in producing the cerebrospinal fluid 1,2 . These tumors account for only 4-8% of gliomas and, after astrocytomas and oligodendrogliomas, ependymomas are the least common 3 . Nearly one-third of brain tumors in patients younger than three years old are ependymomas and constitute around 5%-9% of all neuroepithelial malignancies 1,4 . These tumors are also found in the choroid plexus and may occur at any age, from one month to 81 years and without any gender preference 5 . In pediatric cases, the location of the tumor is intracranial, while adult ependymal tumors can have either an intracranial or a spinal localization 6,7 . The prognosis is better in older children as compared to young infants but nonetheless, in children with intracranial ependymomas, event-free survival after five years is less than 50% 8 . In adults, about 50% to 60% intracranial ependymomas are supratentorial; however, pediatric supratentorial ependymomas account for 25% to 35% of all ependymomas 5,9 . Adults present better prognosis with a 5-year survival of around 90%, while in the pediatric population it is around 60%. The five-year survival rate for supratentorial, infratentorial, and spinal cord ependymomas is 62%, 85%, and 97%, respectively, and for grade I, II, and III spinal cord ependymomas the five-year overall survival rate is 92%, 97% and 58%, respectively 10-12 .
Ependymoma tumors are well circumscribed, soft, tan-red masses and may be associated with hemorrhage. Their microscopic appearance shows hypercellularity and distinct infiltrative margins with surrounding parenchyma, consisting of monomorphic cells with nuclear atypia and brisk mitotic activity. They may also have intramural or glomeruliod vascular proliferation, pseudopalisading necrosis, perivascular pseudo rosettes (5-10% cases), calcifications and hyalinized vessels 1 . Other diagnostic hallmarks include areas of fibrillary and regressive changes such as myxoid degeneration, palisading necrotic areas and the formation of true rosettes, composed of columnar cells arranged around a central lumen 1,6 . Immunologically, they are positive for epithelial membrane antigen (EMA), glial fibrillary acidic protein (GFAP) and S-100. According to the 2016 updated World Health Organization (WHO) classification of brain tumors, ependymomas are divided into 4 types on the basis of histologic appearance: (1) grade I subependymomas, (2) grade I myxopapillary ependymomas, (3) grade II ependymomas, (4) grade II or III RELA fusion-positive ependymomas and grade III anaplastic ependymomas 13,14 .
Previous studies have shown the use of comparative genomic hybridization (CGH) arrays to distinguish intracranial ependymomas from spinal ependymomas 15 . In contrast to other types of brain tumors, histological grading is not a good prognostic marker for outcome for ependymomas 16,17 . Several gene expression studies have been helpful in differentiating between intracranial and extra cranial ependymomas, but have not had clinical significance in directing therapy and their role in tumor origin and prognosis is not clear 18,19 . Studies using cDNA microarrays have shown that gene expression patterns in ependymomas correlate with tumor location, grade and patient age 20 . Cytogenetic studies have shown that chromosomal abnormalities are relatively common in ependymomas 21 . Loss of 22q has been the commonest abnormality found in ependymoma and, in some other tumors, gain of 1q or loss of 6q was observed 21,22 .
To date, there is a lack of information regarding the mutational signatures which distinguish the various subgroups of ependymomas. Another ependymoma cohort study found very few mutations and gene amplifications but a high expression of multi-drug resistance, DNA repair and synthesis enzymes 23 . Intracranial ependymomas differ from spinal ependymomas in the expression of these proteins, and protein expression is also dependent on the ependymoma grade 23 . For both intracranial and spinal ependymomas, very few mutations were reported by using whole exome sequencing 24 . In another study, profiling of NGS mutations was carried out for one case of grade II ependymoma 25 using a GlioSeq panel, which contains a total of 30 genes. In order to determine the mutational patterns of grade III anaplastic ependymoma, we have sequenced DNA from this ependymoma tumor using the Ion Proton system for next generation DNA sequencing with the Ion Torrent's AmpliSeq cancer HotSpot panel. This panel contains 50 genes, only 15 of which also appear in the GlioSeq panel used in previous research. These data provided an evaluation of mutational signatures of this anaplastic ependymoma which differs from the previous two studies, but confirms their conclusions about finding very few mutations in cancer driver genes, helping to direct diagnosis and therapy for ependymomal tumors.

Ethical statement
This study was performed in accordance with the principles of the Declaration of Helsinki. This study was approved by the Institutional Review Board (IRB) bioethics committee of King Abdullah Medical City (KAMC), Makkah, Kingdom of Saudi Arabia (IRB number 14-140). A written informed consent was obtained from the parent of this patient before starting the study.

Clinical specimen
Specimens from all patients willing to give written informed consent and diagnosed with gliomas were eligible to be included in this study. Specimens that cannot be unambiguously identified (no label or specimen number, and/or sample and requisition do not match), specimens with hematomas and blood clots, or specimens from patients who refused inclusion in the study were not eligible. The single patient's tumor tissue (FFPE sections in PCR tubes) used in this NGS analysis was obtained from the histopathology laboratory of Al-Noor Specialty Hospital Makkah, after tumor excision and left frontal craniotomy in the neurosurgery department. The tumor was classified based upon similarity to the constituent cells of the central nervous system, such as astrocytes, oligodendrocytes and ependymal, glial cells, mitosis and cell cycle-specific antigens, used as markers to evaluate proliferation activity and biological behavior (the WHO grading system) 13 . The final diagnosis was made following radiological, histopathological and immunological examinations.

Radiology and histopathological analysis
A CT scan of the brain was performed by a multi-slice CT (MSCT), using a 64-detector-row scanner. The use of computed tomography (CT) allowed visualization of detailed images of the soft tissues in the body in 3D as well as in multiplanar reconstructions. Images were acquired with 5mm slice thickness throughout on a GE Medical Systems, light speed VCT, 64-slice multidetector CT (MDCT). High quality images were processed at low dose performance on Volara™ digital DAS (Data Acquisition System).
The excised tumor was fixed in 4% buffered formaldehyde, routinely processed and paraffin embedded. Four-micrometer-thick sections were prepared on clear ground glass microscope slides with ground edges and routinely stained using Dako Reagent Management System (DakoRMS) with hematoxylin and eosin (H and E) on a Dako Coverstainer (Agilent). For immunohistochemistry, sections were collected on Citoglas adhesion microscope slides (Citotest). Mouse monoclonal beta-catenin (14) (Sigma-Aldrich, cat. no. 224M-1), mouse monoclonal EMA (E29) (Sigma-Aldrich, cat. no. 247M-9), rabbit monoclonal EGFR (SP84) (Cell Marque, cat. no. 414R-16-ASR), mouse monoclonal Vimentin (vim 3B4) (Ventana-Roche, cat. no. 760-2512), GFAP EP672Y rabbit monoclonal  and E-cadherin (36) mouse monoclonal  and mouse monoclonal anti-Ki-67 (Leica Biosystems, cat. no. KI67-MM1-L-CE) antibodies were used for immunohistochemistry. Briefly, the tissue sections were deparaffinized with EZ Prep (Ventana, at 60°C for 1 hr. Immunohistochemistry was performed with the Ventana BenchMark XT automated stainer (Ventana, Tucson, AZ). After inactivation of the endogenous peroxidase using a UV-inhibitor for 4 min at 37°C, the primary antibody was added for 16 min at 37°C, followed by the application of HRP Universal Multimer for 8 min, and detected using the ultraView Universal DAB Detection Kit (cat. no. 760-500) for 38 min. Slides were counterstained with hematoxylin for 8 min and bluing reagent for 4 min before mounting with cover slips. Following staining, images were acquired using NIKON Digital Microscope Camera -DS-Ri1, with image software NIS Elements v.4.0. Appropriate positive controls for all of the studied antibodies were used.
DNA isolation and NGS analysis DNA isolation was carried out using the QIAamp DNA FFPE Kit (50), Cat. No. 56404. 5-10 Formalin-Fixed Paraffin-Embedded sections of 5 microns were deparaffinized using xylene, treated with ethanol to remove the xylene, and the pellet was dried at 65°C for 5 mins. The pellets were resuspended in ATL buffer then treated with proteinase K. The remaining steps were carried out according to the user manuals. DNA concentration was measured using Nanodrop2000C and 10 ng of DNA was used for NGS analysis. DNA was sequenced using the Ion PI v3 Chip Kit (Cat no. A25771, Thermo Fisher Scientific, USA) with the Ion Proton System (Cat no. 4476610, Thermo Fisher Scientific, USA) 26 . Libraries were prepared using Ion AmpliSeq cancer HotSpot Panel v1 (Cat no. 4471262, Thermo Fisher Scientific, USA) primer pools. The Ion AmpliSeq Library Kit 2.0 (Cat no. 4475345, Thermo Fisher Scientific, USA) and Ion PI Hi-Q OT2 200 Kit (Cat no. A26434, Thermo Fisher Scientific, USA) was used for library and template preparation respectively. Sequencing was carried out using Ion PI Hi-Q Sequencing 200 Kit (Cat no. A26433, Thermo Fisher Scientific, USA) reagents and libraries were tagged with Ion Express Barcode Adapters 1-16, Cat. No. 4471250 (Thermo Fisher Scientific, USA). After sequencing, amplicon sequences were aligned to the human reference genome GRCh37 (hg19) (Accession no. GCA_000001405.1) in the target region of the cancer HotSpot panel using the Torrent Suite Software v.5.0.2 (Thermo Fisher Scientific, USA). Variant call format files (vcf files) were generated by running the Torrent Variant Caller Plugin v5.2. Variant calling and creation of vcf files can also be carried out using nonproprietary software such as SAMtools 27 or VarScan2 28 , which also provide coverage analysis. The vcf file data were analyzed using Ion Reporter v5.6 (ThermoFisher Scientific, USA), which calculated allele coverage, allele frequency, allele ratio, variant impact, clinical significance, PolyPhen 2 scores, Phred scored, SIFT scores, Grantham scores and FATHMM scores. This vcf file analysis was also carried out by Advaita Bioinformatics' iVariantGuide. PolyPhen2, SIFT, variant impact and clinical significance can be calculated using non-proprietary software SnpEff 29 and SnpSift 30 . FATHMM scores can also be predicted using fathmm 31 and Grantham scores according to the formula as described in Grantham, 1974 32 . The heat map was generated by the clustering of predicted variant impact scores by Ion Reporter v5.6. The most deleterious score was picked for every gene to generate the heat map; thereafter, hierarchical clustering was conducted. The color codes indicate the following variant impacts using score values 0-8: (0) unknown; (1) synonymous; (2) missense; (3) non-frameshift block substitution; (4) nonframeshift indel; (5) nonsense; (6) stop-loss; (7) frameshift block substitution or indel; (8) splice variant.

Clinical presentation and radiology
A six-year-old female patient presented with a history of right facial palsy for few months with ataxia and right-sided weakness. The patient had a chronic headache, vomiting and had repeatedly been treated for sinusitis. Unenhanced computed tomography (CT) of the brain was performed ( Figure 1, panels A, B and C). A large lesion (5.4 x 7.5cm) was noticed in the left cerebral frontoparietal region. There was an indication of a predominant cystic component and large, eccentric clump of coarse calcification. Additionally, mass effect resulting in midline shift, along with mild scalloping of the internal cortex of the parietal bone, was noted. No hydrocephalic changes or intrinsic hemorrhagic focus were seen ( Figure 1).
Histopathological examination revealed sheets of neoplastic cells with round to oval nuclei and abundant granular chromatin. A variable dense fibrillary background and endothelial proliferation was also noted. Hematoxylin and eosin (H&E) staining results are shown in Figure 2 and Figure 3. Panels A and   Figure 2 show the tumor exhibiting delicate cytoplasmic processes, perivascular rosettes characteristic of ependymoma, focal calcification areas and pseudo palisading necrosis, characterized by a garland-like structure of hypercellular tumor nuclei lining up around irregular foci of tumor necrosis. Panel C shows glomeruloid vascular proliferation and panel D shows extensive palisading necrosis and true rosette formation. The exhibition of a true rosette with a central lumen and the formation of pseudo-palisading necrotic areas is also clear from Figure 3 (panel A). Panel B shows focal areas with numerous tumor giant cells and the presence of brisk mitotic activity, vascular formation and pseudo-palisading necrotic areas. Formation of true rosettes surrounding the microvascular proliferation within ependymal tumors usually signifies anaplastic transformation, which is characteristic of grade III ependymoma (panels C and D). Immunostaining is shown in Figure 4: (A) Ki-67 stain shows a high proliferation index, (B) vimentin positive, (C) GFAP positive, (D) EMA showing punctate cytoplasmic (perinuclear dot-like positivity) staining which is fairly diagnostic of ependymal tumor cells. Figure 5 shows beta-catenin positive (panels A and B) and E-cadherin positive (panels C and D) immunostaining, with both membranous and true rosette-like structures clearly visible in this staining. EGFR staining was negative (see Underlying data) 33 .

NGS data analysis variant identification and variant statistics
Alignment to the target regions (CHP2. 20131001.designed) of the reference genome (hg 19) was performed by the Ion Torrent Suite software v.5.0.2. For this tumor, NGS generated 6,252,341 mapped reads using the Ion PI v3 Chip, with more than 90% reads on target. Amplicon and target base read coverages for the sequencing are shown in Table 1. All 207 amplicons were sequenced with Ion AmpliSeq Cancer HotSpot Panel primer pool. As shown in Table 1, for this sample sequencing the uniformity of amplicon coverage was 95.17%, and the uniformity of base coverage on target was 94.81%. The average reads per amplicon was 34, 179, and the average target base coverage depth was 31,771. 100% of amplicons had at least 500 reads and the percentage of amplicons read end-to-end was 89.37% (Table 1).   Initial analysis by the Ion Reporter 5.6 program found that a total of 1652 variants passed all filters. Initial analysis by Advaita's iVariantGuide software showed 100% (1633) of variants passed all filters (see Extended data) 33 . The filter flags signify variants which do not meet certain criteria during variant calling. The flags refer to the quality or confidence of the variant call. The parameters of flags were read in from the input vcf file. If a variant passes all filters, it is marked as having passed. Six hundred and fifteen variants were identified using a filter for clinical significance that identifies drug response, likely to be pathogenic and pathogenic variants. The distribution of these variants, based on chromosomal position, region within the gene, variant class, functional class, variant impact and clinical significance, are shown in doughnut charts A -F ( Figure 6). As shown in doughnut chart A, chromosome 17 has the highest number of variants (26%) and chromosome 8 has lowest number of variants (0.8%). 98.7% of variants are exonic and, according to variant class distribution, 73.8% are SNPs, 70.2% are missense variants, 25.4% are high impact variants and 46.8% are pathogenic. We have considered true mutations to be those with a Phred score above 20 and significant mutations called by Ion Reporter software were those with a p-value below 0.05.
A summary of the all missense mutations found in the grade III tumor is shown in Table 2. In this tumor, NGS data analysis identified 19 variants, of which four were missense mutations, eight were synonymous mutations and seven were intronic variants. Known missense mutation c.395G>A; p.(Arg132His) in exon 4 of the IDH1 gene, c.1173A>G; p.(Ile391Met) in exon 7 of the PIK3CA gene, c.1416A>T; p.(Gln472His) in exon 11 of the KDR gene and c.215C>G; p.(Pro72Arg) in exon 4 of the TP53 gene were found in this tumor. The frequency, allele coverage, allele ratio, p-value and Phred score for these mutations is shown in Table 3. The p-values and Phred scores were significant for all of these mutations. The frequencies of the three missense mutations, namely PIK3CA c.1173A>G, KDR c.1416A>T and TP53 c.215C>G, were high, suggesting that these are germ line variants, whereas the IDH1 variant frequency was low (4.81%). As shown in Table 2 Table 2). A known splice site mutation (c.1310-3T>C) at an acceptor site in FLT3 (rs2491231) and a single nucleotide variant in the 3'-UTR of the CSF1R gene (rs2066934) were also identified. Additionally, in SMARCB1 a 5'-UTR variant, and an intronic variant in ERBB4 and PIK3CA respectively were found. In Figure 7, the heat map of the variant impact for each gene is presented. The color gradation from green to red indicates unknown, synonymous, missense, nonsense, and splice variants, based upon their SIFT, PolyPhen2 and Grantham scores. Only variants in four genes had a positive PolyPhen2 score (variants in TP53, PIK3CA, IDH1 and KDR genes had a PolyPhen2 score of 0.083, 0.011, 0127 and 0.003, respectively). However, FATHMM scores for the prediction of the functional consequences of a variant suggest that only the IDH1 variant is pathogenic, with a score of 0.94. As described in the COSMIC data base, FATHMM scores above 0.5 are deleterious, but only scores ≥ 0.7 are classified as pathogenic.

Discussion
Ependymomas are brain tumors that arise throughout the central nervous system, within the supratentorial areas, the posterior fossa and the spinal cord. Histologic low-grade (WHO grade I) tumors, such as subependymomas and myxopapillary ependymomas, are usually slow progressing variants of ependymomas. In contrast, grade III ependymomas display anaplastic features like hypercellularity, high mitosis, proliferation of endothelial cells and palisading necrosis 34 . Histopathological evaluation of ependymoma tissue reveals pseudo-rosette formation, high     carcinomas; ERBB4 mutations in lung adenocarcinomas; FGFR3 mutations in breast, endometrial and ovarian cancers; CSF1R mutations in prostate cancer; EGFR mutations in lung adenocarcinomas; RET mutations in thyroid carcinomas; HRAS mutations in melanomas; and SMAD4 mutations in breast cancer. However, with the exception of the KDR variant c.1416A>T, this is the first time the above variants are reported in a brain tumor 49-57 .
We found an intronic variant in PIK3CA and one missense mutation in this gene. This missense mutation was also reported previously in hemangioblastoma and in colon adenocarcinoma 58,59 . Missense mutations in PIK3CA are known to promote glioblastoma tumor progression 60 . Mutations of the PTEN gene are rare in ependymomas and we have also not detected any PTEN mutations in this tumor 61 .
Mutations in cancer driver genes such as TP53, CDKN2A, and EGFR, which are frequently affected in gliomas, have been shown to be rare in ependymomas 43,61,62 . We have detected a TP53 mutation (c.215C>G, p.Pro72Arg, rs1042522) in this tumor with a frequency of 47.94%. This mutation p.(Pro72Arg) has also been reported previously in a medulloblastoma tumor in a young patient 63 . Previous studies have shown that out of 15 ependymoma tumors tested, only one case, a patient with a malignant ependymoma of the posterior fossa, had a mutation in exon 6 of the TP53 gene, which was silent, and in another study only one out of 31 ependymoma tumors tested contained a mutation in the TP53 gene 64,65 . However, in another study, out of 15 ependymoma tumors, none had a mutation in the TP53 gene, suggesting that this gene does not play an important role in the pathogenesis and development of ependymomas, unlike other brain tumor types 61,65,66 . Miller et al., (2018) through whole-exome sequencing of an anaplastic ependymoma tumor, have shown mutations in several cancer-related genes, as well as genes related to metabolism, neuro-developmental disorder, epigenetic modifiers and intracellular signaling 67 . These authors have shown resistance-promoting variant expression in a single ependymoma case at different stages of recurrence. However, these genes were not present in the cancer panel we used in this study. Using the human exome capture on Illumina, Bettegowda et al., (2013) have reported that in one out of eight grade III intracranial ependymomas, tumors have mutations in PTEN and TP53, and one tumor with HIST1H3C mutations 24 . The HIST1H3C p.(Lys27Met) mutation has also been reported previously in posterior fossa ependymomas 68 . Ependymomas may in fact represent a very heterogeneous class of tumors, each with distinct molecular profiles and, even within posterior fossa ependymomas, there are at least two distinct gene expression patterns, as demonstrated by Witt et al., (2011) 19,69 . Overall, in previous studies, a very low frequency of mutations was observed in both intracranial and spinal ependymomas and our findings also supports this observation 19,24,25,41 .
The Ion AmpliSeq Cancer HotSpot Panel consists of 207 primers in 1 tube, targeting 50 oncogenes and tumor suppressor genes that are frequently mutated in several types of cancers. The detected mutations were found to have high accuracy; 100% amplicons had at least 500 reads and 500x target base coverage was also 100%. This high level of accuracy and the high depth of coverage achieved with the Ion Proton system allowed us to reliably detect low frequency mutations with high confidence. Allele coverage in most of the variants is around 2000x, the p-value was 0.00001 and the Phred score was very high for all the variants, indicating high confidence in the variants found in this tumor. Apart from its use in whole-exome sequencing, cancer panel analysis has also become common practice for Ion Proton 26 . The Ion Proton instrument has the advantage of pooling samples using barcodes and the Ion PI chip. For pooled samples, sequencing enables a high throughput up to 15 Gb of data, with more than 60-80 million reads passing read filtering. The purpose of read filtering is to discard the reads that contain low quality sequences, to remove polyclonal reads, remove reads with an off-scale signal, remove reads lacking a sequencing key, remove adapter dimers, and remove short reads etc. If the computed mean read length from all the reads and the minimum total mapped reads in the sample is less than the specified threshold, that sample does not pass the quality control.
Recent molecular diagnostics research had helped in subdividing glioblastomas, oligodendrogliomas and oligoastrocytomas into genetically diverse groups of tumors, and these mutational markers may help in predicting the prognosis and response to therapy 70 . However, such a strategy for the molecular subdivision of ependymomas has been not successful so far using mutational profiling. Epigenetic markers and fusion protein analysis have also helped in identifying new groups of supratentorial ependymoma tumors and in spite of the histopathological signs of malignancy, a small set of ependymomas had a very good prognosis, suggesting that this subgroup of tumors should not be diagnosed as classic ependymomas 71 . However, another study showed that methylation profiling did not identify a consistent molecular class within the supratentorial tumors, but successfully sub-classified posterior fossa ependymoma into two subgroups 72 .
In conclusion, we have identified four known missense mutations, eight synonymous and seven intronic, in this grade III ependymoma. Out of these, only one mutation in FLT3 (c. In this study authors analyzed mutation patterns by next generation sequencing (NGS) in order to determine genomic changes in an anaplastic ependymoma. Authors identified one novel synonymous mutation and one deleterious missense mutation in this grade III ependymoma.

Comments:
In the clinical specimen section authors state "Specimens from all patients willing to give written informed consent and diagnosed with gliomas were eligible to be included in this study." However, the specimen used in this study for NGS analysis was obtained from a single patient's tumor tissue. What are the other samples and in what study they were used? Figure 2A: authors indicate yellow arrow showing palisading necrosis; is this the correct location?
Authors conclude they identified four known missense mutations, eight synonymous and seven intronic, in this grade III ependymoma. How global these signature molecules in the context of grade III ependymoma? Can the authors add additional correlation analysis by analyzing TCGA or other bioinformatics based data?
Authors indicate FLT3 (c.1776T>C, synonymous) is novel, which is interesting; is this mutation reported in other cancer than ependymoma? What is the clinical and functional significance of this mutation in ependymoma progression?
Is the work clearly and accurately presented and does it cite the current literature?
Partly Comment: 2. Figure 2A: Authors indicate yellow arrow showing palisading necrosis; is this the correct location? Answer: As suggested by the reviewer the corrections are made in this figure, and new figure-2 and figure legend will be added at the time of the revision.

Comment:
3. Authors conclude they identified four known missense mutations, eight synonymous and seven intronic, in this grade III ependymoma. How global these signature molecules in the context of grade III ependymoma? Can the authors add additional correlation analysis by analyzing TCGA or other bioinformatics-based data? Answer: As suggested by the reviewer we have searched various databases including TCGA, and summarize our answer to the reviewer's comments as below. A relevant section of this summary will be included in the final revision of the manuscript.
In the TCGA projects ( ) genes such as, TP53, IDH1, KDR, https://www.intogen.org/search?cancer and EGFR in GBM, TP53, IDH1, PIK3CA, and EGFR in Low-Grade Glioma, TP53 and PIK3CA in medulloblastoma are detected as a mutational cancer driver genes. In the NCI's Genomic Data Commons (GDC) portal ( ) 19,144 cases https://portal.gdc.cancer.gov/ are reported in 4 projects of gliomas. Only in Brain projects, for IDH1 6 mutations were found in 423 cases, 369 cases were affected with 239 mutations in TP53; 118 cases were affected with 59 mutations in PIK3CA; 90 mutations in 283 cases were found in EGFR. 0 mutations in 57 cases in HRAS, 38 cases with 20 mutations in RET, and in FGFR3 40 cases with 8 mutations, 32 cases with 15 mutations for ERBB4, in APC 33 mutations in 38 cases, in PDGFRA 110 cases with 45 mutations, in SMAD4 23 cases with 3 mutations, in SMARCB1 24 cases with 6 mutations, in CSF1R 35 cases with 16 mutations, in KDR 102 cases with 28 mutations, in FLT3 63 cases with 20 mutations.
In the TCGA database the missense IDH1 mutation p.(R132H) affected cases are 90.07% (381/423) VEP impact (Ensembl database) is moderate for this, and Shift impact is deleterious low confidence (score 0.01), and the PolyPhen impact is also possibly damaging (score 0.813). In the NCI's Genomic Data Commons (GDC) 423 cases (36.94%) were affected in the IDH1 gene out of 1137 cases. IDH1 c.395G>A p.(R132H) mutation (rs121913500), Clin variant database, #VCV000156444 variant is reported in 2 cases of oligodendroglioma grade II, astrocytoma grade IV respectively (PUBMED ID. 28125199). Also, this variant is reported in 2 cases of anaplastic ependymoma in this database. This variant also reported in AML as an adverse prognostic factor (PUBMED ID. 20368538). IDH1/IDH2 but not TP53 mutations predict prognosis in glioblastoma patients (PUBMED ID. 24868540). The variant we found in the present ependymoma case such as in TP53, HRAS, SMAD4, PIK3CA not in TCGA. However, in the ClinVar database SMAD4 synonymous variant is reported (Accession: VCV000132693.2, Variation ID: 132693).
TP53 mutations in GBM mostly point mutations that lead to a gain of function (GOF) of the oncogenic variants of the p53 protein (Zhang Y, et al., 2018). The TP53 variant rs1042522we reported in the present case [(c.215C>G,p. (Pro72Arg)] was also reported in anaplastic astrocytoma grade-III (Pessoa IA, et al., 2019), but not in ependymoma cases. In the COSMIC database, this variant is reported in several types of cancers. A mutation in Exon 5 of the TP53 gene was reported in one anaplastic ependymoma out of three cases (Tominaga T, et al., 1995). gene was reported in one anaplastic ependymoma out of three cases (Tominaga T, et al., 1995). Whereas in our case the mutation was found in exon 4.
Targeted therapy is being studied for the treatment of childhood ependymoma and other brain tumors utilizing the genomic data. NCI supported Clinical Trials for Ependymomal brain tumors 4 clinical trials are listed ( ). One study is https://www.cancer.gov/types/brain/patient/child-ependymoma-treatment-pdq enrolling the patients for drug targets ( ) Carboplatin and Bevacizumab for Recurrent Ependymoma that inhibit VEGF-promoted angiogenesis. Based on the interesting results observed in the reported small series of patients with recurrent ependymomas treated with bevacizumab, as well as on the evidence of VEGF-promoted angiogenesis in these tumors, we designed a phase II study to test the efficacy of bevacizumab in patients with recurrent ependymoma. Cabozantinib, a multi-kinase inhibitor of FLT3, MET, VEGFR2, and KIT, respectively, and clinical trials are undergoing for Non-Small Cell Lung Cancer, thyroid cancer, AML, and GBM treatment.
Amplification of PDGFRA, VEGFR2 (KDR), and EGFR in gliomas are reported (Puputti et al., 2006), and VEGFR2 plays a key role in neovascularization and tumor initiation by glioma stem-like cells (Yao et al., 2013). We have searched The NCI's Genomic Data Commons (GDC) and found 28 mutations in 102 affected cases with KDR mutations in all TCGA GBM projects. The expression of the KDR gene is increased in endothelial cells during tumor angiogenesis, and missense mutations cause constitutive activation of VEGFR2 in hemangioma. Patients with infantile capillary hemangioma are known to have constitutive activation of VEGFR2 signaling and carry a germline mutation (C482R) in the KDR gene (Jinnin, et al., 2008).
In non-small cell lung cancer patients, a SNP (Q472H), is associated with increased VEGFR2 activity, and it was correlated with increased microvessel density (Glubb DM et al., 2011). This missense mutation is observed in the present ependymoma case also by us, this missense mutation was not known in ependymomas previously. This mutation is reported in colorectal cancer, melanoma, non-small cell lung cancer, and it's an important target for drugs like Avastin (Bevacizumab), Aflibrcept, and drugs reported in the database at ( ). http://atlasgeneticsoncology.org/Genes/GC_KDR.html FLT3 is also expressed in the human brain, though its activating mutations were found mostly in AML . Several insertions, missense, and duplication mutations are known in AML at Val592 codon, for example, FLT3 missense mutation, c.1775T>A p.(V592D) is a pathogenic one in AML. However, in the present ependymoma case, we have observed a mutation in c.1776T>C p. (Val592Val) is a synonymous one. One study reported that this gene is down-regulated and it's associated with favorable clinical outcomes in glioma patients, thus this driver gene might be potential prognostic biomarkers for glioma patients. . In another study, it was reported that two target genes (FLT1, FLT3) of the experimental drug sorafenib were recurrently deleted, whereas another target (KDR) of sorafenib was recurrently amplified in glioblastoma multiforme. (Tran HV, et al., 2018). The FLT3 intronic variant c.1310-3T>C (rs2491231) was identified in 84% of triple-negative breast cancer cases (Uscanga-Perales et al., 2019). This variant was not reported in ependymomas previously, this is the first time we report it here. SMARCB1 mutation in c.1119-41G>A (rs5030613) is reported in Schwannomatosis. This disease is the third major form of neurofibromatosis, clinically and genetically distinct from neurofibromatosis type 1 (NF1) and neurofibromatosis type 2 (NF2). SMARCB1 germline