Alternative lengthening of telomeres (ALT) in pediatric high-grade gliomas can occur without ATRX mutation and is enriched in patients with pathogenic germline mismatch repair (MMR) variants

Abstract

Background: To achieve replicative immortality, most cancers develop a telomere maintenance mechanism, such as reactivation of telomerase or alternative lengthening of telomeres (ALT). There are limited data on the prevalence and clinical significance of ALT in pediatric brain tumors, and ALT-directed therapy is not available.

Methods: We performed C-circle analysis (CCA) on 579 pediatric brain tumors that had corresponding tumor/normal whole genome sequencing through the Open Pediatric Brain Tumor Atlas (OpenPBTA). We detected ALT in 6.9% (n = 40/579) of these tumors and completed additional validation by ultrabright telomeric foci in situ on a subset of these tumors. We used CCA to validate TelomereHunter for computational prediction of ALT status and focus subsequent analyses on pediatric high-grade gliomas (pHGGs) Finally, we examined whether ALT is associated with recurrent somatic or germline alterations.

Results: ALT is common in pHGGs (n = 24/63, 38.1%), but occurs infrequently in other pediatric brain tumors (<3%). Somatic ATRX mutations occur in 50% of ALT+ pHGGs and in 30% of ALT- pHGGs. Rare pathogenic germline variants in mismatch repair (MMR) genes are significantly associated with an increased occurrence of ALT.

Conclusions: We demonstrate that ATRX is mutated in only a subset of ALT+ pHGGs, suggesting other mechanisms of ATRX loss of function or alterations in other genes may be associated with the development of ALT in these patients. We show that germline variants in MMR are associated with the development of ALT in patients with pHGG.

Keywords: ATRX; Telomere; alternative lengthening of telomeres; mismatch repair; pHGG; pediatric brain tumors.

Figure 1.

Alternative lengthening of telomeres (ALT) is more prevalent in pediatric pHGGs than other CNS tumors and can be computationally determined. (A) C-circle analysis was completed for PBTA primary tumor samples from unique pediatric patients (N = 579). Tumor abbreviations: ET, Embryonal tumors, including CNS embryonal tumors and embryonal tumors with multilayered rosettes; GCT, germ cell tumor; GNT, glioneuronal tumor; LGAT, other low-grade astrocytic tumors; JPA, juvenile pilocytic astrocytoma; PXA, Pleomorphic xanthoastrocytoma; SEGA, Subependymal giant cell astrocytoma; GG, ganglioglioma; CP, craniopharyngioma; EP, ependymoma; MB, medulloblastoma; ChP, choroid plexus tumor; ATRT, atypical teratoid rhabdoid tumor; HGG, high-grade glioma. Benign lesions, nonprimary brain tumors, tumors from patients >21 years at time of diagnosis, tumors with fewer than 5 samples per disease type, and duplicate samples for a single patient were excluded from this analysis. pHGG represents 10.8% of tumors analyzed, but 60% of ALT+ tumors. (B) Representation of the pHGG subset (N = 85) on which paired tumor/normal WGS was performed. There was sufficient DNA available to perform C-circle assay on 63 samples and ultrabright telomeric foci (UBTF) analysis was performed on 24 samples. (C) Molecular subtypes in pHGG subset (N = 85). (D) Using the C-circle assay data as the “truth” set for an ALT phenotype, we used the R package cutpointr to determine the optimal tumor/normal telomere content ratio cutpoint for determining ALT +/− status. Shown are density plots for ALT + or − pHGGs at a cutpoint ratio of 1.0679 (x-intercept). (E) This cutpoint enabled a 90.59% accuracy, 93.75% sensitivity, and 88.68% specificity, shown with the receiver operating characteristic (ROC).

Figure 2.

Molecular phenotypes and genomic alterations of pediatric pHGGs by alternative lengthening of telomeres (ALT) status. Annotations for sex (estimated from germline WGS), tumor phase of therapy, TelomereHunter telomere ratio (tumor vs. normal), C-circle assay, ultrabright telomeric foci assay, and immunohistochemistry for ATRX, H3 K28M and H3 K28me3 are shown. TMB is annotated for hypermutant (100 Mut/Mb > TMB ≥ 10 Mut/Mb) and ultra-hypermutant (TMB ≥ 100 Mut/Mb) tumors. Positivity for variants in germline and/or mutations in somatic mismatch repair (MMR) genes (listed in Supplementary Table S3) is annotated above the individual somatic mutations in TP53, H3F3A, ATRX, NF1, TERT, HIST1H3B, and DAXX.

Figure 3.

Alternative lengthening of telomeres (ALT)+ pHGGs are significantly enriched for ATRX mutations and have a higher tumor mutation burden. (A) pHGG patients with ALT are more likely to have ATRX mutations (P < .001, N = 16/32 ALT+, N = 5/53 ALT−). ATRX mutations in ALT+ pHGG are more likely to be likely oncogenic mutations (N = 13/16) compared to mutations in ALT− pHGG (N = 2/5). (B). Mutations in ATRX are significantly associated with ALT (P < .001). (C) ALT+ pHGGs have a higher TMB than ALT− pHGGs (P = .0038). (D) ATRX WT ALT+ tumors have a higher TMB compared to ATRX WT ALT− pHGGs (P = .0098). pHGGs may be ALT positive with (E) or without (F–H) ATRX protein expression in situ. Left and middle panels: Representative images of multiplex immunofluorescence of ultrabright telomeric foci (UBTF) (red), ATRX protein (yellow), or both within DAPI stained nuclei (dark blue) of ALT+ pHGG tissues from 4 patient tumors (E: 7316-158; F: 7316-3058; G: 7316-3765; H: 7316-114). Right panels: representative H&E images and ATRX IHC, noting that in E (tumor 7316-158) ATRX protein expression is absent in tumor nuclei (blue) with positive ATRX staining in non-tumor nuclei. The remaining ALT+ pHGGs (F–H) demonstrate ATRX protein staining.