Germline analysis of an international cohort of pediatric diffuse midline glioma patients

Abstract

Background: Factors that drive the development of diffuse midline gliomas (DMG) are unknown. Our study aimed to determine the prevalence of pathogenic/likely pathogenic (P/LP) germline variants in pediatric patients with DMG.

Methods: We assembled an international cohort of 252 pediatric patients with DMG, including diffuse intrinsic pontine glioma (n = 153), with germline whole genome or whole exome sequencing.

Results: We identified P/LP germline variants in cancer predisposition genes in 7.5% (19/252) of patients. Tumor profiles differed, with the absence of somatic drivers in the PI3K/mTOR pathway in patients with germline P/LP variants versus those without (P = .023). P/LP germline variants were recurrent in homologous recombination (n = 9; BRCA1, BRCA2, PALB2) and Fanconi anemia genes (n = 4). Somatic findings established that the germline variants definitively contributed to tumorigenesis in at least 1% of cases. One patient with recurrent DMG and pathogenic germline variants (BRCA2, FANCE) showed a near-complete radiological response to PARP and immune checkpoint inhibition.

Conclusions: Our study determined the prevalence of pathogenic germline variants in pediatric DMG and suggests a role in tumorigenesis for a subset of patients.

Keywords: PARP inhibitor; diffuse midline glioma; germline variants; homologous recombination; pediatric.

Figure 1.

Germline and somatic landscape of diffuse midline gliomas (DMG) patients with germline deleterious variants. Accompanying somatic pathogenic variants are depicted for 19 patients with germline P/LP variants and DMG. While somatic HR defects were present, only one patient had a second somatic hit in an HR locus (FANCD2) also seen in the germline. There was an absence of PI3K-AKT-mTOR somatic deleterious variants. “HR,” homologous recombination; “MMR,” mismatch repair; “BER,” base excision repair; “MAPK-ERK,” “mitogen-activated protein kinase - extracellular-regulated kinase”; “RTK,” receptor tyrosine signaling. “PI3K-AKT-mTOR,” “phosphatidylinositol-3-kinase - Ak strain transforming- mammalian target of rapamycin”; “BMP,” “bone morphogenetic protein.” “H3K27 var,” H3K27 variant; “Mutant” refers to patients with a known somatic H3K27M mutation who do not have further details regarding the type of histone mutation; “Altered” refers to H3K27 altered without H3K27M mutation

Figure 2.

The germline landscape contributes to somatic copy number aberrations.. Four patients with deleterious germline variants (PALB2 [n = 2]), BRCA2 and FANCE [n = 1], BRCA2 and FANCD2 [n = 1]) and corresponding circos plots of somatic variants are depicted (A–D), including copy number gains (green) and copy number losses (red). There are varying levels of genome instability, with more copy number changes observed in Patients C and D who have 2 deleterious germline variants each. (E) Contribution of somatic TP53 pathogenic/likely pathogenic (P/LP) variants to genome-wide loss of heterozygosity (LOH). Patients with somatic TP53 P/LP variants had a significantly increased genome-wide LOH compared to somatic TP53 wild-type patients (P < .0001). Patients who have an underlying germline P/LP variant in an HR-related gene (termed homologous recombinant deficiency (HRD) in this figure) are depicted as triangles, and colored white for patients harboring multiple P/LP germline HRD mutations. In TP53 wild-type patients, while HRD mutants had higher median genome-wide LOH proportions than HRD intact patients, this trend was not significant at α = 5% (P = .059). “HRD,” homologous recombination deficiency, refers to those patients with elevated mutational signature 3 based on COSMIC v2. “VUS,” variant of unknown significance in BRIP1.

Figure 3.

Radiologic response to PARP inhibition in a diffuse midline glioma (DMG) patient with pathogenic germline variants in BRCA2 and FANCE. T1-weighted MRI brain with contrast in a patient with relapsed DMG, germline BRCA2, and FANCE pathogenic variants, following treatment with a PARP inhibitor and durvalumab at (A) pretreatment, (B) 6 weeks, (C) 2.5 months and (D) 5 months after commencement of therapy. Axial non-contrast T2-weighted images show the same time points in panels E, F, G, and H, respectively.

Figure 4.

Germline variants in DNA damage response are enriched in diffuse midline glioma.. There were 15 variants detected in the DNA damage response pathway: 9 in homologous recombination, 4 in the Fanconi anemia pathway, 2 in mismatch repair. The 3 types of DNA damage response pathways altered in the diffuse midline gliomas cohort are demonstrated with an illustration of the type of DNA lesion that triggers the relevant pathway. The darker blue circles indicate genes with germline pathogenic/likely pathogenic (P/LP) variants detected in our cohort; and the dark blue stars are representative of a single patient with a germline P/LP germline variant. The lighter blue circles are some of the partner genes in the pathway that were unaffected in the cohort.