Fig. 1: Detection of tumorigenic alterations in paired DMG tissue, CSF, and plasma/serum.

a Left: MEC (left y-axis) and Exons 500X (%; right y-axis) of sequencing libraries generated from 30–75 ng DNA input (ID 846CSF). ‘30ng-1’ and ‘30ng-2’ indicate technical replicates of 30 ng inputs. Right: Detection of DMG-associated mutations in genes H3-3A, PPM1D, and PIK3CA from libraries generated at 30, 50, 60, and 75 ng input. b Concordance of DMG-associated mutations detected in paired tumor tissue (T), CSF (C), and plasma (P) or serum (S). ‘X’: variant detected in tumor but not in paired liquid specimen. VAF variant allele frequency. c Representative overlap between tumor-associated mutations (SNVs, insertion/deletions) identified in paired specimens from two patients (IDs 933 and 1446). d Left: Comparison of H3K27M detection by ctDNA sequencing (black) and ddPCR (grey) in paired specimens (n = 8). Right: Detection of H3-3A K27M (10% VAF) by deep sequencing (ID 933CSF), with zero false positive reads of H3C2 K27M. e CNV plots showing chromosomes (chr) 4 and 12 (ID 1549). KIT and PDGFRA gains were detected in tumor and CSF, while KRAS gain was detected exclusively in CSF. X-axis = chromosome position; y-axis = fold-change (FC) calculated based on pre-established baseline; y = 1.3, CNV reporting threshold. f Detection of KIT (n = 1), PDGFRA (n = 1), KRAS (n = 1) and MDM4 (n = 2) gains in tumor and biofluid pairs. y = 1.3: CNV reporting threshold. g DNA methylation array-derived CNV plots of chr1 from pre-treatment (PT) and postmortem (PM) tumor tissue (ID 760), showing gain of chr1q containing MDM4, as validation of MDM4 gain. Y-axis: log-2 CN ratio. T tumor, C CSF, P plasma, S serum, CN copy number.