COMP-10. THE MUTATIONAL PROFILE OF GLIOBLASTOMA-DERIVED CELL-FREE DNA IN PLASMA REPRESENTS A DISTINCT SUBSET OF THE SOMATIC MUTATIONS PRESENT IN GLIOBLASTOMA SOLID TUMOR DNA

Abstract

Abstract Glioblastoma’s mutational landscape varies widely in the same tumor. Using conventional criteria to identify mutations from a focal tissue specimen (e.g., variant allele frequency > 5%) undersamples glioblastoma’s broad clonal diversity, which may limit detection of glioblastoma-derived circulating cell-free DNA in plasma (i.e., circulating tumor DNA; ctDNA). Here, we sought to enhance somatic variant identification in solid tumor DNA to improve detection and characterize glioblastoma-derived ctDNA. Tumor DNA and plasma cell-free DNA (collected < 24 hours prior to the surgical procedure) were isolated from eight glioblastoma patients. DNA was capture-enriched using a custom-designed, glioblastoma-targeted, next-generation sequencing panel (124 genes, 118 kb) followed by paired-end sequencing. Samples were prepared in duplicate with molecular barcodes to enable detection of very-low frequency variants. Somatic mutations in tumor DNA were identified using variable allele frequency thresholds and were considered positive in ctDNA if present in both duplicate samples. Using a lower allele frequency threshold to identify mutations in tumor DNA significantly increased detection of ctDNA (F(1.04,7.29)=14.81, P=0.006). At a solid tumor allele frequency threshold of ≥ 5%, only a single patient (12.5%) had tumor mutations detected in ctDNA. However, at a threshold ≥ 1%, all patients (100%) had at least one tumor mutation detected in ctDNA. Moreover, at a threshold ≥ 0.5%, 7 out of 8 patients (87.5%) had > 12 tumor mutations present in ctDNA. The increased detection of ctDNA enabled the subsequent discovery that somatic mutations in APC, KIT, MSH6, and NF1 were more likely to be present in ctDNA compared to somatic mutations in ATRX, LZTR1, SLC26A3, and TERT which were absent in ctDNA (χ 2=8.0, P=0.005). Thus, stronger sampling of glioblastoma’s genetic heterogeneity in tumor DNA improves detection of ctDNA allowing comparisons between mutational profiles that may lead to the identification of similarities and differences with key biologic and/or clinical implications.