Diagnostic evaluation of long-read nanopore sequencing for pediatric CNS tumors.
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
Long-read nanopore sequencing enables methylation-based CNS tumor classification while providing additional genomic information within a single assay. However, its performance across molecular features relevant to pediatric CNS tumor diagnostics and across available methylation classifiers remains incompletely characterized. We evaluated nanopore sequencing for integrated molecular characterization of pediatric CNS tumors and compared with data from DNA methylation arrays and Illumina short-read sequencing used in the clinical diagnostic work-up. Twenty tumor samples representing three major pediatric CNS tumor groups (gliomas, ependymomas and medulloblastomas) were subjected to long-read nanopore sequencing, which generated robust long-read data across all tumor groups, with a mean genome coverage of approximately 40x. Nanopore-derived copy-number profiles showed good concordance with methylation array-derived profiles (median Pearson r = 0.816). Nanopore evidence was identified for 74/78 (95%) clinically relevant small variants detected by Illumina sequencing, although 53/78 (68%) were recovered as exact high-confidence nanopore calls. Structural variant profiles showed limited breakpoint overlap and differed in the number, type, and size distribution of detected events between Illumina and Nanopore sequencing. Nanopore-derived methylation classifications were largely concordant with integrated WHO CNS5 diagnoses. Ependymomas, medulloblastomas, and pilocytic astrocytomas, were concordantly classified across all evaluated classifiers, although not all classifications reached the respective classifier-specific confidence thresholds. In contrast, diffuse gliomas showed greater variability in classification across classifiers. Downsampling indicated that concordance with integrated WHO CNS5 diagnoses remained high at substantially reduced sequencing depths, whereas the proportion of classifications reaching classifier-specific confidence thresholds was more classifier dependent. Long-read nanopore sequencing enables integrated molecular characterization of pediatric CNS tumors within a single assay, although concordance with established methods varied by genomic feature and tumor type. The preservation of methylation classification performance at reduced sequencing depth supports the feasibility of shorter sequencing workflows. Prospective studies are required to establish performance and turnaround time in routine clinical diagnostics.