Biom-19. Methylation Profiling Of Plasma Cell-Free Dna In Pediatric Brain Tumor Patients

Shejuan An,Kathleen McCortney,Jordain Walshon,Kaethe Leonard,Michael DeCuypere,Craig Horbinski
DOI: https://doi.org/10.1093/neuonc/noae165.0092
2024-11-29
Neuro-Oncology
Abstract:Circulating tumor DNA (ctDNA) assays are being evaluated to inform clinical decisions in cancer care in adult patients, but the study in pediatric brain tumor patients is rare partly due to very low levels of ctDNA in blood. Currently, treatment decisions require invasive diagnostic surgical biopsies that carry risks and morbidity. The aim of this study is to utilize methylomes from plasma ctDNA for non-invasive diagnosis in pediatric brain tumor patients. Using cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq), which with low input DNA requirement, we measured methylation profiles of plasma samples from 77 pediatric brain tumor patients and 16 patients with non-neoplastic diseases like epilepsy. Binomial GLMnet classifiers of tumor or tumor subtype were built, in 2000 iterations of 60% training sets. Performance was evaluated in 40% test sets. RESULTS. These 77 pediatric brain tumor patients were a mixture of the most common pediatric brain tumors, include circumscribed astrocytic gliomas (N=23), glioneuronal tumors (N=16), embryonal tumor (N=9), ependymal tumor (N=7), pediatric type diffuse high grade glioma (N= 5), glioma NOS (N=4), mesenchymal tumor (N=4) and other pediatric brain tumors (N=9). The methylation profile differentiated brain tumors from non-neoplastic diseases with 82.9% accuracy (precision = 92.6%, sensitivity = 86.2%, specificity = 66.7 %). For subtype analysis, the methylation profile detected circumscribed astrocytic glioma from non-neoplastic diseases with 85.7% accuracy (precision = 87.5%, sensitivity = 87.5 %, specificity = 83.3 %), and glioneuronal tumors from non- neoplastic diseases with 83.3% accuracy (all precision, sensitivity, and specificity = 83.3 %). The results suggest that methylation profiling of plasma cell-free DNA has the potential to discriminate between pediatric brain tumor patients and patients with non- neoplastic diseases. Larger population cohorts to train even more accurate classifiers will be needed.
oncology,clinical neurology
What problem does this paper attempt to address?