The germline variants in DNA repair genes in pediatric medulloblastoma: a challenge for current therapeutic strategies

Brain tumors represent the leading cause of childhood cancer mortality. The most common malignant brain tumor among them is medulloblastoma [1]. Although multimodality treatment regimens have substantially improved survival in this disease, up to 30-40% of patients with medulloblastoma still die of the disease. Detrimental effect of current treatment on long-term survivors is also observed [2]. Our understanding of the molecular background of pediatric brain tumors has expanded significantly over the past few years. The vast amount of genomic and molecular data generated recently has proved that medulloblastoma is not a single entity but is composed of at least four subtypes: Wingless (WNT), Sonic Hedgehog (SHH), Group 3 and Group 4 (non-WNT/SHH types), with distinct genetic and biological profiles as well as different course of disease requiring adequate therapeutic approaches [2–6]. Despite of improved understanding of the molecular basis of medulloblastoma, many cases still lack an obvious genetic driver [4, 7, 8]. The further research focused on additional potential mechanisms responsible for the development of this tumor may lead to identification of new susceptibility factors as well as new markers that predict response to therapeutic agents and provide prognostic information. So far, majority of driver mutations detected in medulloblastoma are of somatic character. Impact of germline genetic variability that may affect clinicopathologic presentation of this tumor have not been in-depth investigated yet [5, 7, 9–12].

In our study we focused on evaluation of germline defects in genes that play a role in DNA repair pathway because of the following reasons. Firstly, DNA-repair deficiency is associated with cancer development and the key role of germline alterations in promoting tumorigenesis is highlighted by several cancer predisposition syndromes e.g. Li-Fraumeni, Fanconi anemia or Turcot syndrome, where occurrence of medulloblastoma has been recorded. Secondly, it is well known that germline defects may modulate the response to treatment since DNA-repair mechanisms make cells prone to the effects of DNA-damaging chemotherapy [13–15]. It is important to notice that majority of evidence about the impact of DNA-repair genes defects on toxicity in medulloblastoma comes from either description of single cases [16, 17] or from mouse models and cell lines experiments [13, 18] but not from systemic clinical based investigation. Therefore, all these data indicate that DNA repair genes are a promising targets possibly linked both to development of tumor and response to therapy in medulloblastoma.

Within essential components of DNA repair signaling cascade the NBN gene particularly draws attention as potentially susceptibility marker for medulloblastoma [19, 20]. Germline defects in medulloblastoma patients were observed also in other genes cooperated with NBN in BRCA1-associated genome surveillance complex (BASC), including MSH6, PMS2 and MLH1 [21–24].

Biallelic defects in NBN gene result in Nijmegen Breakage Syndrome (NBS; OMIM:251,260), while homozygous defects in MSH6, PMS2 or MLH1 genes are molecular cause of Constitutional Mismatch Repair Deficiency Syndrome (CMRDS; OMIM:276,300), hereditary disorder associated with increased risk of cancers including medulloblastoma [25]. Among other genes responsible for CMRDS is also MSH2 (ID:4436, MIM:609,309), one of the key factor of DNA mismatch repair system which recognizes and repairs mispaired or unpaired nucleotides resulted from DNA replication errors [25]. There is an evidence that germline MSH2 defects may predispose to primary early-onset CNS tumors, especially glioblastoma [26]. In addition, De Rosa et al. suggest that in some families with Turcot syndrome the coexistence of colorectal and childhood brain tumors may result from a complete MMR deficiency [27]. However, association between MSH2 defects and medulloblastoma was not evaluated yet.

A very similar phenotype to NBS was seen in patients with Nijmegen Breakage Syndrome-like Disorder (NBSLD – OMIM:613,078) caused by defects in the RAD50 gene (ID:10,111, MIM:604,040). This gene encodes the protein involved in DNA double-strand break repair, cell cycle checkpoint activation, telomere maintenance and meiotic recombination suggesting that molecular variants disrupting its function may lead to genome instability and carcinogenesis [28]. Furthermore, inactivation of proteins like RAD50 required for the homologous recombination machinery leads to defects in the nervous system development indicating that components of this system can play crucial role in development and progression of various neuro-oncological diseases [29]. The frequency of the molecular variants in RAD50 gene was, similarly to MSH2, not determined in medulloblastoma patients up to now.

Therefore the first purpose of this study was to establish the spectrum of germline defects in MSH2 and RAD50 genes, as well as frequency of two known NBN variants in 102 patients with medulloblastoma. In the next step we have evaluated the hypothesis that DNA repair genes may affect a response to therapy in medulloblastoma patients. We have found that alterations in a range of DNA repair genes are associated with occurrence of rare severe adverse effects during chemotherapy in patients.

Assuming that the fundamental feature of cancer is genomic instability, functional defects of proteins which are responsible for maintenance of genome integrity by correcting DNA replication errors, should be carcinogenic. It is therefore not surprising that a number of cancer susceptibility genes encode key factors of DNA repair pathways. Recent comprehensive analysis of germline mutations in pediatric cancers pointed to DNA repair genes as the most commonly mutated genes, including TP53 and BRCA2 [43]. It is also increasingly clear that defects in DNA repair genes may determine patient’s response to radio and chemotherapy [13, 16, 17]. In view of that we evaluated the potential association between DNA repair defects and treatment related toxicity as well as their potential role as a susceptibility factor for medulloblastoma.

The sequence analysis of two well-known repair genes MSH2 and RAD50 conducted in large cohort of 102 medulloblastoma patients revealed three new germline variants MSH2 p.V606I and p.A733T as well as RAD50 p.R1093*. Both the localization and the character of detected variants allow for prediction of their probably pathogenic impact on the encoded proteins what was supported by the results of the in silico analysis (Table 2). The p.V606I and p.A733T substitutions are localized in the crucial DNA mismatch repair protein V (MutSV - aa 619-854, pF00488) in highly (p.V606I- phyloP:4.40) and moderate (p.A733T-phyloP:2.55) conserved amino acid region. MutSV domain contains the dimerization interface and nucleotide-binding site with C-terminal helix-U-turn-helix motif that is critical for MutS function [44]. The RAD50 p.R1093* variant resulting in premature stop codon has severe consequences on the protein translation and predicts suppression of its protein. All identified variants were uncommon in our patients (1/102; 0.98%). This is consistent with the published data indicating that molecular variants in MSH2 and RAD50 in CNS tumors occurred very rarely (5/1637 – 0.31% and 4/1743 – 0.23%, respectively [37]. In support of that, in published recently study of germline mutations in pediatric cancers, including medulloblastoma, MSH2 and RAD50 variants were not reported [43]. Due to the lack or low frequency of candidate variants in control groups the estimation of cancer risk associated with their presence was not possible (Table 2). However, deleterious character of detected germline variants, the role of the encoded proteins in DNA repair system and their annotation with genetic syndromes, including NBSLD and CMRDS associated with medulloblastoma, make them the potential susceptibility variants for this kind of tumor. The RAD50 p.R1093* variant was reported as one of two known molecular defects (HGMD CMO92910) responsible for NBSLD. In medulloblastoma patients pathogenic variants in MLH1, MSH6 and PMS2 genes were detected previously [21–24, 43]. Alterations in these genes together with MSH2 defects lead to CMRDS. Both MSH2 and RAD50 encode the crucial components of the DNA repair system. MSH2 belongs to mismatch repair genes (MMR) while RAD50 together with MRE11 and NBN constitute the MRN complex responsible for connecting DNA damage detection to DNA repair and cell cycle checkpoint function [44, 45]. Biallelic deficiency in MMR genes had been referred as a molecular cause of increased predisposition to gastrointestinal and hematological malignances, as well as early-onset CNS tumors (especially glioblastoma; GBM) [22, 26]. Additionally, the germline heterozygous variants in MMR gene were reported in patients with Turcot syndrome associated with medulloblastoma incidence. The molecular variants affecting genes of the MRN complex might also play a role in pediatric tumor development. The evidence that NBN heterozygous variants predispose to childhood acute lymphoblastic leukemia and medulloblastoma was already published [19, 46–48]. All these facts reinforce potential role of DNA repair genes, including MSH2 and RAD50 in susceptibility to medulloblastoma but detailed mechanistic studies are required to confirm this preliminary hypothesis.

The MRN complex genes, including MRE11, NBN and RAD50 are required for double-stand DNA break (DBS) repair via one of the DNA repair system, homologous recombination (HR). Defects in HR system lead to hypersensitivity to agents that produce DSB and topoisomerase inhibitors eg. etoposide [14, 15, 49]. MMR genes remove mispaired nucleotides by the cooperation in mismatch repair system whose defects are associated with hypersensitivity to DNA crosslinks and platinum-based chemotherapeutic agents (eg. mitomycin C and carboplatin) [14, 15, 49].

Medulloblastoma treatment protocol (Additional file 2: Figure S1 A and B) includes the drugs mentioned above, specifically platinum-based chemotherapeutic agents (carboplatin, cisplatin), topoisomerase inhibitor etoposide and, in addition, mitotic inhibitor vincristine. Therefore it is very likely that our patients with molecular variants in DNA repair genes may be more prone to complications in recovering from chemotherapy induced DNA damage. They include the patients with variants detected by WES analysis in ERCC2, FANCM or EXO1 genes, an essential components of DNA repair systems [14]. All three identified variants (p.R695C, p.L694* and p.V738L, respectively) were localized in highly conserved nucleotide position (phyloP: 0.89-0.99) in crucial for the encoded protein domains (Table 2) and the character of detected variants (nonsense and splice site) strengthens their pathogenic role. Previous functional studies had also linked these variants to increased sensitivity to therapeutic agents. Defects of ERCC2 protein were reported as a cause of faults in the nucleotide excision repair mechanism (NER) which is responsible for removal of variety of helix-distorting DNA lesions, as well as a hypersensitivity to platinum derivatives. The FANCM gene is one of the elements of the Fanconi Anemia (FANC) pathway responsible for DNA crosslinks repair, possibly through coordination of three main DNA repair systems: nonhomologus endjoing (NHEJ), homologus recombination (HR) and translesion DNA synthesis (TLS). Loss of function of this system results in sensitivity to DNA crosslinking agents and platinum derivatives [14, 15, 49]. Finally, EXO1 gene encode a nuclease which cooperates with MRN complex in DSB repair via HR pathway as well as interacts with MMR genes in repair of DNA mismatches [50, 51].

Although we acknowledge that functional studies are necessary to explore the mechanism through which DNA repair gene defects influence the treatment related toxicity we have already found significant association between defects in NBN, MSH2, RAD50, FANCM, ERCC2 and EXO1 genes and clinical data. Indeed, more than half of patients with variants in DNA repair genes suffered from rare adverse grade 4 events after administration of chemotherapy (Table 3). We acknowledge that validation cohorts would be necessary for confirmation of our results. Unfortunately, recently published NSG ‘discovery sets’ of medulloblastoma ranged only from 39 to 92 samples and molecular defect in MSH2, RAD50 and NBN gene were not identified [7, 9–12, 43]. Also, an information related to the therapy and accompanied side effects was not provided in these studies. However in two pediatric medulloblastoma patients with mutations in DNA repair genes (PALB2 and BRCA2) chemotherapy inducted grade 4 side effects were reported [16, 17]. In addition, effect of other drugs being introduced to medulloblastoma treatment protocols e.g. temozolomide (TZM) may be dependent on the status of mismatch repair genes. In melanoma one variant g.73170T>C in MSH2 gene (rs2303428) was associated with response and side effects and could be used as a molecular marker for TMZ treatment response [52].

On the other hand it is difficult to compare toxic effects caused by cancer treatment in adult patients harboring defects in DNA repair genes with toxicity observed in still developing and vulnerable tissues in children. Different spectrum of tumors in children and therefore different treatment protocols, including very high doses of drugs, may influence dissimilar reaction in the latter population.