Recognizing and framing germline variants related to a cancer predisposition is an evolving topic in the field of cancer genomics. In 10% of pediatric cancers, pathogenic germline variants are detected in genes related to oncogenesis or progression of cancer, suggesting that some individuals are predisposed to develop certain oncological diseases (11–13). The onset and progression of cancer can therefore be modulated by the interaction and crosstalk between germline and somatic variants (14).
High-throughput NGS testing is increasingly being performed in children with cancer to identify pathogenic variants in cancer predisposition genes even in cases where the clinical features are not highly suggestive of an underlined CPS, leading to a diagnosis that is essential for accessing cancer surveillance and screening in family members (12, 13).
Patients suffering from CNS malignancies are overall more likely to have an underlying genetic predisposition than patients with other malignancies (11, 12, 15, 16) and there is growing evidence from analyses of international multicenter cohorts, including retrospective and prospective clinical studies and patient series, that additional genes predispose to pediatric brain cancer (16–19). Traditionally, genetic testing is recommended when patients present with a) particular tumor type, b) significant family history, c) synchronous or metachronous tumors, d) unexpected treatment toxicity, e) features consistent with a specific CPS, f) somatic genetic findings suggestive for a constitutional variant (20). With the emergence of genome-wide approaches, there has been a rise in the occurrence of occasional identification of germline alterations predisposing to cancers (21), suggesting a greater extent of CPS than previously identified. Defining risk models that predict the probability of specific genetic variants predisposing to HGG is one of the most ambitious projects (22).
In this study, cancer-predisposing genes were investigated in 77 patients who were originally diagnosed with HGG, and germline variants were identified in 44 cases (57.1%). In our cohort, gene variants associated with CPS were reached in 13 patients (16.9%), a higher rate compared to the literature on the incidence of CPS in children with any malignancy (1, 13, 23).
Muskens et al. evaluated the presence of CPG variants in a group of 280 pediatric patients including those with HGG, detecting the presence of mutations favouring oncogenesis in 31 patients (11.1%) and showing the high recurrence of pathogenic variants, especially in the TP53 gene (13). This trend was in line with our findings, where TP53 germline LP variants were found in 2 of our patients. In P10, the rs1060501191 has paternal inheritance and the variant was found also in the patient’s somatic sample (AF of 96%) (24); rare cancer types are reported in our patient’s family members, so the family oncological history was highly consistent with the diagnosis of Li-Fraumeni Syndrome (MIM #151623). On the other hand, the result of the inheritance analysis was not known in P11 since the parents refused to perform the test, in addition P11 has a grandmother who developed a CNS tumor.
In addition, we identified only one patient (P1) with CMMRD syndrome, affected by astrocytoma IDH-mutant and an earlier-onset Burkitt's lymphoma, in whom a homozygous variant was found in MSH6 gene. CMMRD syndromes are characterized by the development of primary CNS tumors and multiple colon adenomas with a high risk of progression to adenocarcinoma. Compound heterozygosity or homozygosity mutations in the MLH1, MSH2, MSH6, or PMS2 genes are associated with an increased risk of the onset of HGG, the leading cause of death in these patients (3). In P1, the same MSH6 variant was found in simplex heterozygosity in the parents and two siblings; therefore, this genetic information has allowed them to be diagnosed with Lynch syndrome (LS) (OMIM #614350). In P2, P3, and P4, we detected pathogenic variants in three CPG not linked in scientific literature to the pediatric CNS tumor phenotype found in our study. In addition, in P5 a DMG H3 K27-altered was diagnosed, a rarely finding associated with pathogenic variant of NF1 (25).
In addition, among the subjects with a likely pathogenic variant, P8 was diagnosed with an infant-type hemispheric glioma and the heterozygous c.65T > C (p.Leu22Ser) variant in the BRCA1 gene. BRCA1 (BReast CAncer gene 1) is an onco-suppressor gene associated with dominant familial breast and ovarian cancer (MIM#604370) in autosomal dominant form and Fanconi's Anemia (MIM#113705) in autosomal recessive form. People carrying a single pathogenic variant in the BRCA1 gene have an increased risk of developing cancer. To date, an association between mutations in the BRCA1 gene and an increased risk of CNS malignancies has not been described; however, a possible role as a contributing factor cannot be excluded. Of note, in P9 a likely pathogenic c.910dupG in the POT1 gene with paternal inheritance that is neither described in the literature, nor reported in ClinVar and GnomAD was found. Variants in this gene are associated with the condition known as POT1-TPD and are characterized by an increased lifetime risk of developing multiple cutaneous melanomas, chronic lymphocytic leukemia, angiosarcoma, and gliomas (26). Our patient was affected by high-grade astrocytoma with piloid features onset at the age of 12 years and a kidney cell carcinoma onset at the age of 18 years, a botryoid rhabdomyosarcoma of the bladder and a pediatric brain tumor was reported in two second-degree relatives on the paternal side, indicating a heterogeneous clinical spectrum of this condition (personal communication).
Other variants classified as likely pathogenic, but not reported in the literature, GenomAD and ClinVar were found in patients 7 and 12. In P7 (DMG), the c.1679_1680insT variant in the CDH1 gene with paternal inheritance was found. Pathogenic variants of the CDH1 gene are associated with a hereditary diffuse gastric tumor Syndrome (MIM #137215) with autosomal dominant transmission and have been described in families with inherited forms of glioma (27). In P12 (Neuroepithelial tumor, PATZ1 fusion-positive) a rs768092215 in the ERBB2 gene with maternal inheritance was identified. Variants in ERBB2 are associated with non-BRCA1/2 families of breast cancer and with an increased risk of myeloproliferative neoplasms (28, 29). Although involvement in glioma pathogenesis is not certain, we know that pathogenic variants in the ERBB2 gene are associated with glioblastoma in somatic form (MIM: #137800).
Thus, given the 13 LP/P variants detected in the 77 patients, we can assume a rate of variants related to CPS in 16.9% of our population. This frequency is higher than expected (20) and allows us to speculate to consider performing NGS genetic tests in all HGG patients.
Moreover, VUSs were detected in 62.5% of cases, and the genes more frequently involved were: ATM (4), EGFR (4), and NF1 (3). Multiple VUSs were detected in 7 patients. Globally, 10 variants originally considered VUS variants were reclassified as LB/B. VUSs found in the somatic and germinal samples did not help reclassify them. Two cases are worth mentioning and it might be interesting to study the variants found with functional studies. P19 (Diffuse pediatric-type HGG, H3-wt and IDH-wt) carried a VUS c.1272_1273insGT in the RAD51B gene that was inherited from the mother. The variant was neither described in the scientific literature, nor annotated on ClinVar, and not found in GenomAD, but the patient has a positive family oncologic history (a cousin on the maternal line deceased at age 22 because of a brain tumor). P39 (Astrocytoma IDH-mutant) carried heterozygous VUSs in NF1 (rs786201902), RAD54L (c.160C > G), and ATM (rs138526014), and an LB variant in EPCAM (c.696C > G,), all variants follow paternal inheritance. VUSs in NF1 and RAD54L are not described in the scientific literature, are not annotated on ClinVar, and are not found in GenomAD. In contrast, the variant in ATM is annotated on ClinVar as VUS in Hereditary CPS and is reported associated with an increased risk of breast cancer (30, 31). In addition, the father is in good health. The twin inherited the same variants and recovered from acute lymphoblastic leukemia at age three.
The distribution of variants across different histological subtypes was not different, but pathogenic variants were detected mostly in patients with DMG. Regarding survival in our cohort, we observed an expected lower 12-month OS from after diagnosis in DMG (68.7%) as compared to pHGG (72.6%), and other histological variants (92.9%). Although there was no difference in long-term survival among different variants, patients with P or LP variants have shorter 12-month OS (59.2%) compared to VUS or LB/B/NO (74.1% and 80.9% respectively). To note, patients with P/LP variants affected by DMG presented a 12-month OS of 50%, representing a lower value in comparison to children who carried VUS (70.6%) or LB/B/NO (73%). The same trend was confirmed in the pHGG group: OS at 12 months for P/LP is 50% versus 81.8% and 66.7% for LB/B/NO and VUS, respectively.
Our study is the largest case series describing germline genetic profile in pediatric high-grade CNS tumors, a definite diagnosis of CPS being performed in 16.9% of patients, greater than has been described to date (20). Beyond the known indication for investigate a CPS (20), in our work we show that variants were found even in the absence of specific warning criteria.
Moreover, we recorded the worst outcome for the patient affected by P/LP variants for both DMG and pHGG subgroups, confirming the importance of detecting these variants to provide appropriate family counselling, including aspects on prognosis/survival expectancy.
The study has some limitations to draw definitive conclusions on the impact of germline variants on prognosis and survival especially in diseases with per se dismal prognosis: this is a retrospective study and the series is relatively small and heterogeneous, as the vast majority of the cases included were diagnosed as HGG before 2021 WHO Classification of CNS tumours. Furthermore, the gene panels used are heterogeneous and somewhat restricted to a limited number of genes. Finally, genetics is a rapidly evolving field; consequently, we must focus specifically on the increased application of germline genetic testing because a large amount of data obtained expand understanding of the risk factors associated with pediatric cancer development and they can inform clinical management in the future.