Feasibility of long-read nanopore sequencing for methylation-based classification of posterior fossa ependymomas

Posterior fossa group B (PFB) ependymoma in children poses diagnostic challenges, particularly when tissue is limited or histological features overlap with other entities. Unlike posterior fossa group A tumors, which show EZHIP overexpression and global H3K27me3 loss, recurrent single-nucleotide variants or other coding alterations in PFB remain largely unidentified, highlighting the importance of epigenetic profiling. We applied long-read nanopore sequencing to four pediatric posterior fossa tumors initially diagnosed as PFB ependymoma, including three specimens archived for more than a decade. DNA methylation profiles were analyzed using t-distributed stochastic neighbor embedding (t-SNE) and classified with the Epignostix CNS Tumor Methylation Classifier (v12.8), with comparison to matched Illumina EPIC array data. Cases 1 and 2 were classified as PFB, while Case 3 localized between the PFB and subependymoma clusters. Case 4 clustered with low-grade gliomas on t-SNE analysis, and long-read sequencing identified BRAF p.V600E mutation, excluding ependymoma. Nanopore- and array-based t-SNE embeddings showed concordant clustering across all cases. Long-read nanopore sequencing enabled reliable methylation-based classification from long-term archived tissue and allowed concurrent detection of methylation patterns and genomic alterations with a reduced turnaround time. These results support its feasibility as a complementary approach for integrated molecular diagnosis of central nervous system tumors.