Search

BACKGROUND Constitutional mismatch repair deficiency syndrome (CMMRD) is a severe cancer predisposition syndrome resulting in early onset central nervous system (CNS) and other cancers. International guidelines for surveillance exist but no study has systematically evaluated the efficacy of this protocol. METHODS We surveyed all confirmed CMMRD patients in the International Replication Repair Deficiency Consortium. A surveillance protocol consisting of frequent biochemical, endoscopic and imaging (CNS and total body MRI) studies were employed. Survival analyses and efficacy of each method were assessed. RESULTS Surveillance data were collected from 105 CMMRD individuals from 41 countries. Of the 193 malignant tumors, CNS malignancies were the most common (44%). The surveillance protocol uncovered 49 asymptomatic tumors including 16 glioblastomas and medulloblastomas. Five-year overall survival was 89% for tumors discovered by surveillance, and 61% for symptomatic tumors (p6 months as per protocol. Finally, of the low grade tumors identified asymptomatically, 5 were low grade gliomas. All of the low grade gliomas, which were not resected transformed to high grade tumors at a median of 1.6 ± 0.9 years. CONCLUSION These data support a survival benefit in CMMRD patients undergoing a surveillance protocol. Adherence to protocol and resection of lower grade lesions may improve survival for patients with CNS tumors.

Pediatric brain tumors with replication repair deficiency (RRD) are hypermutant and may respond favorably to immune checkpoint inhibition (ICI). We are collecting ongoing clinical and molecular data from patients with RRD hypermutant cancers treated with ICI as a part of our consortium registry study. Companion biomarkers include tumor mutational burden (TMB), neoantigens and genetic signatures obtained from whole genome and exome sequencing. Immune inference is obtained from RNAseq. Additionally, T-cell receptor rearrangement data are collected from the tumor and blood throughout treatment. From 2015–2018, 53 patients were treated with ICI and combination therapies. Of these 39 had brain tumors, with 93% having high grade gliomas. Two-year overall survival for the entire cohort is 47+/-8%, which compares favorably with historical controls. Tumor location had major impact on outcome. Non-CNS solid tumor patients have 2-year OS of 78+/-11%, all failures occurring within the first 2 months, and sustained responses are observed for 3 years. In contrast, OS for brain tumors is 39+/-10% with late recurrences observed even after 2 years of therapy (p=0.02). Large tumor size and total tumor burden are associated with higher rates of “flare” and poor outcome throughout all cancers. While all tumors are hypermutant, TMB and predicted neoantigens do not correlate with response. However, specific signatures extracted from SNVs and indels are significantly associated with response to ICI and favorable outcome (p=0.005). Notably, RRD IDH1 mutant glioblastomas have particularly poor response to ICI. Interestingly, glioblastomas (n=8) which failed single agent ICI had favorable responses to combination immunotherapies with prolonged survival of 65%+/-8% one year after failure vs 0 for untreated patients (p=0.01). The favorable outcome and responses to ICI are encouraging. Since brain tumors respond less favorably to ICI than other solid tumors, combination therapies based on tumor and immune signatures of these cancers may be beneficial.

Children with pediatric gliomas harboring BRAF V600E mutation have a poor outcome with current chemoradiation strategies. Our aim was to study the role of targeted BRAF inhibition in these tumors. We collected clinical, imaging, molecular and outcome information from BRAF V600E glioma patients treated with BRAFi across 29 centers from multiple countries. Sixty-seven patients were treated with BRAFi (56 pediatric low grade gliomas, PLGG and 11 pediatric high grade gliomas, PHGG) for up to 5.6 years. Objective responses were observed in 80% of PLGGs compared to 28% with conventional chemotherapy (p

Pediatric brain tumors with replication repair deficiency (RRD) are hypermutant and may respond to immune checkpoint inhibition (ICI). We performed a consortium registry study of ICI in recurrent RRD cancers. Clinical and companion biomarkers were collected longitudinally on all patients. Biomarkers included tumor mutational burden (TMB), neoantigens and genetic signatures obtained from whole genome and exome sequencing. Immune inference was obtained by RNAseq and T cell rearrangement was collected in the tumor and in blood throughout treatment. Of the 46 tumors on the study, 32 were brain tumors with glioblastoma in 96%. Rapid, objective responses (>50%) were observed in 50% of glioblastomas. Three year overall survival for the whole cohort was 48+/-8% which compares favorably with historical controls. Brain tumors fared worse with OS of 39+/-10% and late recurrences observed even after 2 years of therapy (p=0.02). Tumor size and acute “flare” constitute poor outcome throughout all cancers. While all tumors are hypermutant, TMB and predicted neoantigens correlated with response to ICI (p=0.02). Specific signatures extracted from SNVs and total mutations predicted response to ICI and favorable outcome (p=0.005). RNA inference and TCR reveal that the FLARE phenotype is mostly acute nonspecific immune response and not true progression. Finally, glioblastomas (n=8) which failed single agent ICI had favorable responses to combinational immunotherapies with prolonged survival of 65%+/-8% at one year after failure vs 0 for other patients (p=0.01). RRD glioblastomas exhibit favorable outcome and responses to ICI. Combinational therapies based on tumor and immune signatures of these cancers are necessary.

BACKGROUND: Hypermutation constitute an important subgroup of cancers and confers sensitivity to immune checkpoint inhibition (ICI). Glioblastoma arising in children with Constitutional Mismatch Repair Deficiency Syndrome (CMMRD) are ultrahypermutant. Our objective was to quantify the frequency of hypermutant tumors in children, determine whether mutational signatures can predict germline mutations, and treat hypermutant tumors with ICIs. METHODS: Deep panel sequencing of 2984 pediatric tumors (585 brain tumors). Tumor mutation burden was correlated to mutation burden from exome and genome sequencing (R(2) = 0.94). Mutational signatures were analyzed to predict source of hypermutation. Data on 36 patients with hypermutant tumors identified by sequencing were enrolled on an ICI registry trial. RESULTS: Hypermutant tumors (>10 mut/MB) comprised 5% of all pediatric tumors (n=143). These were highly enriched for replication repair mutations (p

BACKGROUND: The leading cause of cancer-related mortality among children is brain tumours and glioblastoma multiforme (GBM) has the worst prognosis. Epigenetic deregulation is believed to be a driver as these patients have few mutations compared to adults. New treatments are urgently needed but few pre-clinical in vivo models exist. One approach of generating these is to transplant tumour tissue into mice, but it yields highly variable results and requires serial passaging in mice, which is time-consuming, expensive and ethically questionable. We therefore aimed to establish a cell line-based orthotopic mouse model representative of the patient tumour. METHODS: Patient-derived cancer stem cell (CSC) lines from paediatric GBM were orthotopically transplanted into immunodeficient mice. The xenograft tumours were evaluated by histology for glioma features. Genome-wide DNA methylation analysis was performed on the CSC, xenograft and patient tumours. RESULTS: All CSC lines initiated tumours and the survival of the mice correlated with the survival of their respective patient. The xenograft tumours had key glioma features and were classified as GBM by histology and methylation profiling. The methylation profile of the xenograft tumour clustered together with the patient tumour (

BACKGROUND: A significant amount of childhood brain tumors emerge from cancer predisposition syndromes. Although tumor sequencing is common practice, it is currently impossible to infer germline mutations from tumor data. Glioblastoma Multiforme arising in children with Biallelic Mismatch Repair Deficiency Syndrome (bMMRD) are ultrahypermutant which is highly specific to this syndrome and confers eligibility for immune checkpoint inhibition therapy (ICI). Our objective was to quantify the frequency of hypermutant tumors in children, determine whether mutational signatures can predict germline mutations, and treat hypermutant tumors with ICIs. METHODS: Deep panel sequencing (1000x) of 315 genes in 2984 pediatric tumors (585 brain tumors) and 79 842 adult tumors (3910 brain tumors) was performed. An algorithm was devised to determine mutation burden from 2 MB of panel sequencing data and results correlated strongly with mutation burden from exome sequencing (R2 = 0.94). Mutational signatures were analyzed to predict source of hypermutation. Patients with hypermutant brain tumors identified by panel sequencing were treated with ICI. RESULTS: Hypermutant tumors (>10 mut/MB) comprised 5% of all pediatric tumors (n=143). These were highly enriched for replication repair mutations (p=100 Mut/MB harbored MMR/polymerase mutations suggesting germline bMMRD (p =10–7). Clinical data collected on 15 ultrahypermutant tumors revealed germline mutations in replication repair genes in all patients. Of the 22 patients with hypermutant cancers treated with ICI, 16 had brain tumors, and favorable sustained responses are observed. CONCLUSION: High mutation burden is a sensitive predictor of germline bMMRD. Hypermutant tumors are more common in the pediatric setting than previously appreciated, opening novel therapeutic avenues. ICI shows promise for hypermutant pediatric cancers.