Medulloblastoma (MB) is among the most common malignant brain tumors in children. Despite advances in multi-modal patient care and understanding of tumor biology, one-third of affected children still succumb to their disease. Efforts to improve patient outcome have been hindered by the lack of sensitive biomarkers to stratify treatment response and predict relapse. In relapsing patients, potential tumor evolution with treatment implies that capturing genomic profiles at recurrence might be required for identifying actionable therapeutic vulnerabilities. There is thus a dire need for novel, sensitive biomarker-driven assays to (i) complement conventional imaging-based disease evaluation and (ii) inform molecular targets at relapse. Liquid biopsies have recently shown promise for detecting and tracking tumor-specific genomic alterations, including targeted sequencing of tumor-derived cell-free DNA (cfDNA) collected from the cerebrospinal fluid (CSF) of brain tumor patients. However, the utility of cfDNA analysis for children with MB remains understudied. We recently devised and an experimental pipeline for inferring somatic copy number variants (CNVs) in CSF-derived cfDNA based on low-coverage WGS (lcWGS). In pilot studies, we demonstrated feasibility of inferring genome-wide CNVs based on lcWGS generated from sub-nanogram cfDNA inputs. Tumor-associated CNVs were detected in 70% of cfDNA samples obtained at baseline, with higher detectability in MB patients with metastatic disease than in those without. Analysis of serial cfDNA samples from CSF during treatment and follow-up indicated an association between CNV detectability and disease course. Re-emergence of somatic CNVs was observed in patients who progressed, identifiable >3 months before relapse was diagnosed radiographically. In addition, divergent CNVs were observed in selected patient-matched primary and relapse pairs. Based on these findings, we hypothesize that the detection of tumor-specific somatic alterations in CSF-derived cfDNA will correlate with patient outcomes in an expanded cohort of children with MB, and longitudinal profiling of such will enhance understanding of mechanisms underlying tumor evolution and recurrence. To test these hypotheses, we propose to (i) establish the utility of CSF-derived cfDNA profiling for correlation with disease burden and prediction of progression in a derivation cohort of prospectively treated children with MB; (ii) validate the algorithm of cfDNA analysis in an independent MB trial cohort; and (iii) investigate tumor evolution in MB through comparison of somatic alterations in longitudinal cfDNA samples. These studies will be conducted in an unprecedented cohort of serial CSF samples collected from children enrolled on two prospective, multi-institutional MB trials (n=140 patients; n>600 CSF samples). The proposed research is anticipated to establish the use of CSF as a minimally invasive, sensitive, and robust form of routine liquid biopsy for MB patients with the potential of revolutionizing risk stratification, disease monitoring, and intervention for affected children.
Medulloblastoma (MB) is a highly aggressive childhood brain tumor from which one third of affected patients fail current therapy, relapse, and succumb to their disease. Current standard-of-care for monitoring disease status involves the use of neuro-imaging at routine intervals, which suffers from a general lack of sensitivity to accurately detect microscopic residual disease or early recurrence, hampering opportunities for timely therapeutic intervention. This proposal aims to evaluate and validate the utility of applying genomics to cerebrospinal fluid-derived cell-free DNA collected longitudinally from a large collection of prospective MB trial patients as a novel, sensitive, biomarker-driven assay for the real-time assessment of disease burden and early detection of relapse in affected children.
