PATH-18. CSF-DERIVED CIRCULATING TUMOR DNA REFLECTS DISEASE COURSE AND CLONAL EVOLUTION IN MEDULLOBLASTOMA

Abstract

Abstract Medulloblastoma (MB) is the most common malignant brain tumor in young children. A multi-modal approach comprises standard of care, and recent advances in molecular studies have identified clinically relevant subgroups. Nonetheless, 30% of MB patients relapse or progress, conferring dismal prognosis. Such suboptimal outcome is partly due to a lack of sensitive biomarkers for response-adapted personalization of treatment and relapse prediction. Circulating tumor DNA has been shown to correlate with disease status in a variety of adult cancers, but studies in pediatric brain tumors, including medulloblastoma, are scarce. Here, we aim to evaluate the utility of profiling cell-free DNA (cfDNA) derived from longitudinally-banked cerebrospinal fluid (CSF) samples collected from children with MB enrolled in two prospective, multi-center trials (SJMB03 and SJMB12, estimated sample size of 100 patients, including 50 with subsequent progression). cfDNA was extracted, quantified, and analyzed for size distribution from pre-centrifuged CSF serially banked during the course of treatment and follow-up. Low-pass whole genome sequencing (LP-WGS) enabled detection of chromosomal and focal copy number alterations (CNAs). CNAs detected in cfDNA were compared against known somatic changes in corresponding primary tumors. Detectability of tumor-specific CNAs in cfDNA was then correlated with tumor burden and patient outcome. Pilot analysis showed presence of arm level and/or focal CNAs in cfDNA from 80% of CSF samples in 20 children with metastatic MB, and longitudinal assessment revealed correlation with clinical course throughout treatment with radiotherapy and chemotherapy. Further comparison of tumor and longitudinal cfDNA derived CNAs revealed divergent genomic alterations implicating potential oncogenic mechanisms underlying treatment failure and recurrence. Our study performed on a large prospective series of MB trial patients substantiates the utility of CSF derived cfDNA as an actionable biomarker in high-risk MB patients while also facilitating understanding of tumor evolution and heterogeneity.