Introduction
Choroid plexus tumors (CPT) have arisen as papillary neoplasms of choroid plexus in the brain, particularly in children lateral ventricle or in adult fourth ventricle (1). These tumors originate from the epithelium of choroid plexus. They can be benign or malignant and known as choroid plexus papilloma (CPP) or choroid plexus carcinoma (CPC), respectively (2). In children, CPP account for 2–3% of all intracranial tumors, with a median age at diagnosis around 3.5 years old (3) and an annual incidence of 0.3 cases per million. This neoplasm outnumbers CPC by a factor of 5:1 (4). The tumor is commonly located in the lateral ventricle, and the main symptom is an elevated intracranial pressure. Histopathological characteristics observed in CPP are essentially calcification, hemorrhage, hyalinization, oncocytic changes, and other presenting symptoms, including headache, diplopia and ataxia from hydrocephalus (5). Direct mechanical obstruction of cerebrospinal fluid (CSF) flow, hemorrhage that causes arachnoid granulation obstruction, and CSF overproduction can stimulate the development of hydrocephalus, and occasionally lead to CSF leaks out from the nose (6). Metastasis of CPP tumors exists; however, extensive metastasis has only been reported when tumors are malignant (7). Histologically, CPP corresponds to grade I, according to the World Health Organization (WHO); however atypical CPP (a-CPP) become more aggressive and reach grade II. It very rarely transforms into grade III-CPC (4). Although CPPs are commonly benign, histological analysis is not always predictive of their behavior. Molecular and genetics analyses may therefore be helpful for diagnosis and tumor prognosis (7,8).
Increased prevalence of CPC due to R337H mutation of the germline gene tumor protein p53 (TP53) in southern Brazil was reported (9). Custodio et al (9) demonstrated for the first time that this mutation is responsible for 63% of CPC cases in children, which indicates a higher incidence of CPC in southern Brazil. This study including 22 children with CPC and 7 children with CPP was based on PCR-RFLP and confirmed by sequencing exon 10 of TP53. They reported that all CPP cases (7/7) were negative for R337H (9). A recent study using whole exome sequencing demonstrated that homologous recombination deficiency, amplification of mesenchymal-epithelial transition (MET) and loss of retinoblastoma susceptibility gene were observed in two cases of CPC (malignant peripheral nerve sheath tumor sheath tumor) (10). Conversely with other types of brain tumor, extensive mutation profiling in CPT cases were not previously attempted by next generation DNA sequencing (NGS) methods. However, gene expression profiles using DNA micro-arrays, Sanger sequencing of TP53 gene, copy number variation and DNA methylation studies in CPT were attempted (11,12). It has been reported that mutations in hSNF5/INI1 are associated with CPCs, and one mutation in the SNF5/INI1 gene in one case of a-CPP is known (13). However, Mueller et al (14) have reported that there is no evidence of hSNF5/INI1 point mutations by SSCP analysis in CPP. Furthermore, hSNF5/INI1 (SMARCB1) gene serves an integral role in chromatin remodeling (15). SMARCB1 is an invariant of the SWI/SNF chromatin-remodeling complex and serves an important role in transcriptional regulation (16). Patients with rhabdoid tumors, CPC, epithelioid sarcoma and peripheral primitive neuroectodermal tumor exhibit a loss of SMARCB1 expression either by exons duplications or deletions, although the number of cases with SMARCB1 mutations are limited (17).
Mutations in ATM gene are associated with breast cancer and several forms of leukemia and lymphomas (18). However, no association between the development of brain tumors and ATM gene mutations has been yet established. A previous study demonstrated an association between astrocytoma development and familial ataxia-telangiectasia (AT) in an 8-year-old patient (19). A previous study using Sanger sequencing reported a compound heterozygote for two novel mutations: c.3291delC (p.Phe1097fs) at exon 25 and c.8198A>C (p.Gln2733Pro) at exon 58 in the ATM gene of patient with AT and with a cerebellar astrocytoma (20). Another study using next-generation targeted exome sequencing revealed a frameshift mutation p. L2493fs c.4477_4478CTCT in a patient with astroblastoma, and a missense mutation p. (Val410Ala) in the ATM gene in a patient with astrocytoma from the same group (21). To the best of our knowledge, molecular genetics investigations, particularly ATM gene mutations in patients with CPP and CPC by DNA sequencing using NGS methods have not yet been investigated, and the mutation profiling of these tumors remains unclear. The present study was the first to report NGS analysis of an a-CPP tumor on Ion Proton using cancer comprehensive panel genes and identification of a novel and known mutations.
Discussion
The a-CPP grade II tumor is situated between the CPP and CPC tumors (11). A-CPP tumors were previously considered to be a separate pathologic subgroup that exhibits an intermediate degree of mitotic activity and overall survival rate. The intermediate position of a-CPP between CPP and CPC was supported by the clinical data, such as the presence of metastases at diagnosis, event-free survival rates, and increases in Ki-67 proliferation marker and the p53 tumor suppressor protein across the three histological subtypes (from CPP to a-CPP and then CPC) (26). Histopathological examination of this tumor revealed increased mitotic index of 2 mitoses/10 HPF and cribriform morphology, which confirmed that, according to the diagnostic criteria of WHO classification of CNS tumors, the present case was an a-CPP (27). Extensive metastasis from CPPs can take place into the subarachnoid space when the tumors are malignant (28). In low-grade CPP, the unique papillary configuration is a distinct feature, whereas in CPC the papillary features are unclear and this ill-defined growth pattern sometimes makes the diagnosis challenging (29). The fibrovascular papillary projections of CPP are lined by cuboidal to columnar epithelium compared with normal choroid plexus tissue, which has orderly cobblestone appearance (1,5). CPP tumours commonly exhibit higher nuclear-to-cytoplasmic ratio and nuclear hyperchromasia. This was the first case of a-CPP reported in Saudi Arabia. However, a case of CPP was reported by Jamjoom et al (30).
Most of the CPP cases have strong immunoreactivity to S-100 protein, whereas glial fibrillary acidic protein (GFAP) are expressed weakly and only focally in a few tumors (31). In the present study, no EMA staining was observed, although it has been reported to be commonly positive in CPP cases (32). In the present study, GFAP and EGFR staining was negative, whereas synaptophysin was focally positive. GFAP, synaptophysin, transthyretin and S100 protein stainings were reported to be more variable in CPP cases (33). However, pan-cytokeratin and vimentin are commonly positively expressed in CPP tumours (1,33). It has been reported that most CPPs express cytokeratins, vimentin and podoplanin, and are also usually positive for E-cadherin (4,34). The loss of E-cadherin expression would increase cell proliferation migration and invasion (35,36). Although E-cadherin is present on the basolateral surface of the majority of CPPs, decreased expression in atypical papillomas and CPCs has been reported; however, the present study demonstrated positive staining for E-cadherin (37,38).
The novelty of the mutations identified in the present study (Table I) were verified by COSMIC and ExAc databases. Several molecular pathways associated with CPC have been previously reported and allow distinction of CPC from CPTs (11,39,40). Mutations in TP53 were demonstrated to be associated with increased tumor aggressiveness and worsen survival outcome in CPCs; however, no mutation in TP53 was reported in a-CPP and it is in agreement with the results from the present study (11,39–41). No mutation profiling in CPP cases has been previously attempted by NGS methods; however, it was done in numerous types of brain tumor and used to distinguish these tumors from their mutational profiles (42,43). For example, the distinction between adamantinomatous craniopharyngiomas and papillary craniopharyngiomas was based upon the presence of mutations in the genes catenin-β 1 (CTNNB1) and B-Raf proto-oncogene, serine/threonine kinase following whole exome sequencing (44). To the best of our knowledge, mutations in the genes NOTCH1, ATM, STK36, MAGI1, GATA2, FGFR3, DST, RECQL4, NUMA1, THBS1, MYH11, MALT1, CDH20 and KDM5C have been reported in CPP tumors. STK36 is a serine/threonine protein kinase that serves a crucial role in the sonic hedgehog pathway by regulating the activity of the GLI transcription factors and that is essential for postnatal development, which may be mediated by CSF homeostatis or ciliary function regulation (45). Overproduction of CSF, growth retardation, hydrocephalus development and early mortality were reported in STK36 knockout mice (45). CSF overproduction also worsens the risk of developing hydrocephalus (6). The patient from the present study also presented hydrocephalous and CSF accumulation. This indicates that STK36 mutations may serve a crucial role in choroid plexus tumor progression. However, further investigation is required to understand this phenomenon. Shajani-Yi et al (46) reported that 28% patients with glioblastoma present mutations in TP53, following NGS using Cancer HotSpot Pane, and also identified mutations in clinically actionable genes, including IDH1, IDH2, CDKN2A and PIK3CA (46). In the present study we have also found a known missense mutation in exon 5 of TP53, mutation in c.419C>T p. (T140I COSM43742), with a p. value of 0.05396, Phred quality score of 12.6795 and frequency of 1.36% (not shown in results). A missense mutation in the exon 10 of IDH2 was also observed in this tumor. This mutation in c.1261G>A (p. Gly421Ser COSM3420710) had a frequency of 5.06%, a P-value of 0.01377 and a Phred quality score 18.6103 (not shown in results). These two mutations have Phred score <20, and were subsequently not included in the analyses or Tables. In addition, the tumor from the present study presented a novel synonymous variant in IDH2. The a-CPP tumor assessed in the present study harbored mutations only in the EGFR, a driver gene for glioma, activation and autophosphorylation domains, whereas other driver genes for various gliomas, such as TP53, PTEN, NF1 and IDH1, did not have any mutations (47,48). The present study was the first to reveal a complete mutational profile of a-CPP tumor by NGS technology on an Ion Proton platform and using 409 tumor suppressor and oncogenes targeted for mutations. It has been reported that glioblastoma tumors frequently overexpress the mesenchymal-epithelial transition (MET) proto-oncogene, of which expression correlates with tumor grade. In addition, MET signaling confers resistance to radiotherapy, and various MET inhibition strategies are being developed to treat GBM tumors (49). A novel nonsense mutation c.525T>A p. (Cys175Ter) in a-CPP tumor was reported in the present study.
The present study identified a novel intronic variant in hSNF5/INI1 (SMARCB1) gene. No exonic mutation was found in SMARCB1; however, one missense mutation was detected in SMARCA4 gene. Hasselblatt et al (2011) reported that a nonsense mutation inactivates SMARCA4 in an atypical teratoid/rhabdoid tumor with intact SMARCB1 (INI1) expression (50). To the best of our knowledge, the present study is the first to report that SMARCA4 mutation is associated with a-CPP. For the first time in brain tumors, four missense variants in ATM, GATA2, FGFR3 and KDM5C and one deletion variant in NOTCH1 were identified in the present study. The ATM missense mutation [c.5948A>G; p. (Asn1983Ser)] identified in the present study was of germline origin, according to ClinVar database by ‘Ambry Genetics’ with the variation ID No. 140790. This mutation was also found in SNP database with rs # 659243, and in a patient with familial history gastrointestinal cancer; however, to the best of our knowledge, this mutation has not been reported in patients with brain cancer (24). This ATM mutation [c.5948A>G; p. (Asn1983Ser)] is adjacent to the Ser 1981 and Ser 1985, which is a part of the focal adhesion targeting (FAT) domain that is required for ATM autophosphorylation (51,52). However, this mutation was reported to be benign with a FATHMM prediction score 0.01 (53). The mutation reported in the a-CPP case from the present study in exon 16 of the FGFR3 gene [COSM6854620 c.2044G>A p. (Val682Ile)] had also been reported previously in large intestine (25), with a FATHMM prediction score of 0.97, which suggests that this mutation might be pathogenic (25). However, this mutation has not yet been reported in brain tumors.
The formation of various types of tumor in humans is modulated by NOTCH signaling pathway, which has been reported to serve a crucial role in the development and formation of normal choroid plexus (54). Insertion of G and A in PARP1 gene (c.1056_1057insG and c.1056_1057insA) causes a frameshift in the reading frame and a termination at 35th coding position from this insertion. In addition, insertion of G (c.21904_21905insG) in PARP1 causes a frameshift in the reading frame and a termination codon at 3rd coding position from insertion (55). PARP enzymes are involved in DNA damage repair (56). Mutations in SYNE1 (ARCA1) gene are known to cause autosomal recessive cerebellar ataxia type-1 (57), and patients with CPP are known to have symptoms of ataxia (5). Previous studies reported that SYNE1 and STK36 mutations are present in GBM cases (58). In addition, deletion variant in NOTCH1 [c.4732_4734delGTG p. (Val1578delVal)], which causes a frameshift in reading frame, was also observed in hematopoietic and lymphoid tissues, but this variant has not previously reported in GBM (59). Mutations in NOTCH1 are associated with altered drug sensitivity, such as to tyrosine kinase inhibitors (60). Frameshift-deletion in exon 26 of NOTCH1 has been reported in hematopoietic and lymphoid skin, salivary gland and thymus (61). KDM5C, also known as ‘Jumonji. AT-rich interactive domain 1c’ JARID1C, codes for a histone demethylase, and mutations in this gene are associated with mental retardation and microcephaly (62). This suggests that alteration in chromatin remodeling might serves a role in CPP. It is therefore crucial to further investigate choroid plexus tumor to better understand the role of chromatin remodeling, since this type of tumor exhibits mutations in enzymes responsible for chromatin remodeling, including PARP, SMARCA4 and ATM. A mutation in GATA2 gene [c.967C>T p. (His323Tyr)] has been previously identified in large intestine with a FATHMM prediction score of 1.00 (25). To the best of our knowledge, the missense mutation detected in the GATA2 gene in the present study has not yet been reported in GBM tumors. GATA2 is a hematopoietic factor that has been implicated in hematopoietic malignancies, and this transcription factor has been implicated in prostate tumorigenesis (63,64).
ATM kinase is an important tumor suppressor, and its activation causes the phosphorylation of numerous targets, including proteins involved in cell cycle arrest, DNA repair and apoptosis (65). Blake et al (66) have demonstrated that inactivation of the ATMIN/ATM pathway protects against glioblastoma pathogenesis. Eich et al (67) revealed that ATM and ATR inhibition sensitizes cancer cells to temozolomide (TMZ) and to other anticancer compounds. In addition, several studies also reported that ATM- and ATR (ataxia telangiectasia and Rad3 related)-mutated cells are sensitive to TMZ mediated cytotoxicity (68), Inhibition of ATM-AMPK enhances TMZ-induced cytotoxicity in inherently TMZ-sensitive glioma cells, and ATM kinase inhibition also sensitizes p53 mutant glioma cells to ionizing radiation (69,70). The novel mutation revealed in the present study was located on chromosome 11, in the exon 39 of ATM gene. In addition, amongst patients with prostate cancer in AT heterozygotes, a higher proportion ATM exon 39 polymorphism (G5557A) was reported (71). Copy number analysis by molecular inversion probe method revealed focal chromosomal gains in the region of chromosome 11, which is where ATM gene locus is present (31).
Ion AmpliSeq comprehensive cancer panel consists of 16,000 primers supplied in 4 tubes (aliquots), which targets 409 genes. This panel targets the exons of numerous tumor suppressor genes and oncogenes that are frequently mutated in various types of cancer. The mutations detected in the present study with the Ion Proton system presented high accuracy and high depth of coverage, which allowed to reliably detect low frequency mutations with high confidence. Individual allele coverage for this novel ATM mutation A>T are in forward strands, the sequencing reads for the A allele are 1,026 and 904, respectively, and for the T allele, the sequencing reads are 45 and 24 respectively. This variant coverage was 1999, mutation frequency was 3.45%, allele frequency was 0.034517 and P-value was 0.00407. Instead of using whole exome sequencing, cancer panels analysis also became a common practice on Ion Proton. A previous study analyzed circulating tumor DNA in non-small cell lung cancer by Ion Proton (22). This instrument offers the advantage of pooling more samples using the barcodes, which is what the authors of the aforementioned study did by using Ion AmpliSeq Colon and Lung Cancer Research Panel v2 on Ion Proton. Pooled samples sequencing using Ion PI chip enables a highest throughput up to 15 Gb data, with >60–80 million reads passing filter.
It is crucial to obtain well-calibrated quality scores, since SNP and genotype calling at a specific position in the genome depends on the base calls and the per-base quality scores of the reads overlapping the position. Coverage values represent the read depth (DP), which represents the number of reads passing quality control that are used to calculate the genotype at a specific site and in a specific sample with higher values for DP, generally leading to more accurate genotype calls (72). In the present study, the Phred quality score of the novel ATM variant was 23.909. A score in this range corresponds to a <0.5% error rate in base calling. These results suggested that this novel mutation may be observable in more CPP tumors. The Polyphen score indicates whether a substitution is damaging, and predicts the possible impact of an amino acid substitution on the structure and function of a human protein. Variants with Polyphen scores in the range of 0.0–0.15 are predicted to be benign, with the score of 0.15–1.0 potentially having a damaging effect, and 0.85–1.0 score established as damaging (23). The novel ATM variant revealed in the present study had a Polyphen score of 0.097, which suggested that it may be benign.
Numerous ATM mutants possess the ability to autophosphorylate on Ser-1981 (73). This activation by the autophosphorylation on Ser1981 leads to dissociation of the inactive ATM dimer (or higher-order multimer) into single protein molecules with kinase activity (51). Most missense mutations occur in the C-terminal part of the amino acid 1966. Mutations of the N-terminal part are therefore less likely to result in loss of kinase activity (74). The novel mutation found in this study was Leu1936Phe that was between the autophosphorylation sites viz, Ser1893 and Ser1981 (73). This result suggested that the kinase activity of ATM protein may not be affected by this mutation, since the kinase domain in ATM was situated between the amino acids 2712 and 2962. Also FRAP/ATM/TRRAP (FAT) domain is not affected because it is at region 1960 to 2566 amino acids (51).
In conclusion, the present study identified a novel ATM mutation and 11 novel mutations in an a-CPP tumor, which had not yet been reported in any databases. Since these novel mutations have not been previously implicated in choroid plexus tumors, the findings from the present study may reveal a genetic signature that could be useful to distinguish choroid plexus tumors between CPP, a-CPP and CPC. In addition, the present study was the first reported case of a-CPP from Saudi Arabia. In this tumor, EGFR mutations were detected in the kinase and autophosphorylation domains of EGFR. The presence of mutations in this driver gene may indicate the transformation of a a-CPP grade II tumor to a CPC grade III tumor. Furthermore, the present study revealed mutations in KDM5C, PARP, SMARCA4 and ATM genes, which suggested that they may serve crucial role in chromatin remodeling, and may serve in understanding the progression of the benign tumor to malignant tumor. The results from this study may help the development of targeted therapies for this pediatric tumor, which had not been previously due to the lack of mutational information. This investigation may allow understanding this rare tumor at a molecular level.