Concordance of rapid whole genome and targeted long read sequencing of pediatric brain tumors with standard clinical testing.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
Background: The 2021 WHO classification of central nervous system tumors requires integrated molecular testing for accurate diagnosis and risk stratification. Current clinical workflows typically require separate assays for single nucleotide variants (SNVs), copy number variants (CNVs), structural variants (SVs), and DNA methylation profiling, leading to diagnostic delays of weeks to months and substantial tissue consumption. Long-read sequencing (LRS) offers the potential to deliver all required molecular data from a single assay with dramatically reduced turnaround time. Methods: We evaluated 23 pediatric brain tumor samples, representing 10 different tumor types, using LRS on the Oxford Nanopore Technologies (ONT) platform, performing whole-genome sequencing (wgLRS; n =9), adaptive sampling (AS; n =20), or both ( n =6). Using a streamlined protocol with an approximately 60-minute sample preparation time, we compared LRS-derived variant calls and methylation-based tumor classification against clinical testing results. Results: Both wgLRS and AS showed high concordance with clinical testing across all variant classes. SNVs and indels were detected with high sensitivity above 15% allele fraction (42 of 51 clinically reported variants), and a somatic-optimized caller recovered four additional driver mutations missed by the standard germline pipeline. CNV profiles showed strong agreement with clinical testing, and 5 of 6 clinically reported gene fusions were identified. Methylation classification was concordant with clinical diagnosis in 18 of 22 (82%) evaluable tumors after only 10 minutes of sequencing using Sturgeon alone, increasing to 19 of 22 (86%) within one hour of sequencing using three real-time classifiers. Adding MNP-Flex, which draws on an expanded 184-class reference, resolved one additional tumor for an overall methylation concordance of 20 of 22 (91%). All discordances were attributable to differences in sampled tissue (autopsy versus surgical specimen), low tumor content, subclonal variants below detection thresholds, or limitations of classifier reference sets. Conclusion: LRS allows for comprehensive molecular characterization of pediatric brain tumors from a single library preparation, with methylation-based classification often available within an hour of sequencing initiation. This integrated approach has the potential to accelerate diagnostic workflows, reduce tissue requirements, tailor neurosurgical approaches, and facilitate timely treatment decisions in pediatric neuro-oncology.