Advancing pediatric germ cell tumor classification through nanopore-based transcriptome analysis | LC 25

Biography

Ana Flavia is a PhD student at Barretos Cancer Hospital, Brazil, and also a visiting scholar at The University of North Carolina at Chapel Hill. Ana’s current research in molecular oncology focuses on unraveling novel diagnostic tools for pediatric cancers, specifically on germ cell tumors.

Abstract

Germ cell tumors (GCTs) are a rare and heterogeneous group of neoplasms derived from primordial germ cells, accounting for 3.5% of pediatric tumor diagnoses. Advances in molecular techniques have enabled detailed investigation of the genetic characteristics of GCTs, revealing unique gene expression patterns in pediatric patients through various sequencing platforms. In this study, we utilized nanopore whole-transcriptome sequencing to analyze 56 GCT samples, including fresh-frozen and formalin-fixed paraffin-embedded (FFPE)-preserved tissues. The cohort included 14 dysgerminomas, 13 yolk sac tumors, 11 mature teratomas, seven immature teratomas, eight mixed tumors, and three embryonal carcinomas. Despite the shallow sequencing depth, nanopore sequencing proved to be a robust and cost-effective method, successfully identifying gene expression patterns that distinguished GCT subtypes with high accuracy using a machine learning model. The classifier achieved an overall accuracy of 93.7% with a prediction probability threshold of 0.8. The model demonstrated strong performance for most subtypes, with high F1-scores (0.90 for dysgerminomas, 0.92 for yolk sac tumors, 0.95 for teratomas, and 0.40 for embryonal carcinomas) and near-perfect area under the curve values (0.99 for dysgerminomas, 0.97 for yolk sac tumors, 0.99 for teratomas, and 0.98 for embryonal carcinomas). However, the embryonal carcinoma subtype presented classification challenges due to its low sample representation. These findings highlight the potential of nanopore sequencing as an accessible tool for characterizing pediatric GCTs. Expanding nanopore-based gene expression studies to include a wider variety of GCT subtypes could enhance machine learning models for GCT classification and support broader solid tumor diagnostic efforts.