Abstract PR013: Childhood Cancer Data Initiative’s Research Characterization Data Reveals Novel Biology

Abstract

The Childhood Cancer Data Initiative (CCDI), a 10-year $50M per year data initiative, brings together doctors, researchers, and families affected by cancer to learn from every child diagnosed with cancer, speeding discovery for new medical advances. In partnership with the Children’s Oncology Group (COG), NCI launched the Molecular Characterization Initiative (MCI), a pediatric precision oncology program comprising two arms, clinical sequencing and research characterization. The MCI provides comprehensive molecular profiling for children, adolescents, and young adults with newly diagnosed high-risk and rare pediatric cancers. Clinical testing includes paired tumor-normal whole-exome sequencing, targeted RNA fusion detection, and, for central nervous system (CNS) tumors, DNA methylation profiling to support precision treatment decisions. To extend the value of these biospecimens, the Research Molecular Characterization (RMC) arm performs comprehensive, state-of-the-art, multi-omic profiling on residual specimens and releases the resulting data as a public resource through the CCDI Hub. Baseline research characterization includes paired tumor-normal deep whole-genome sequencing (WGS) and total RNA sequencing for most clinically sequenced cohorts, with selected cohorts undergoing additional cutting-edge analyses, e.g., single-cell RNA-seq, spatial transcriptomics, long-read DNA/RNA sequencing, cell surface proteomics, phosphoproteomics, and metabolomics. These datasets complement the clinical testing reports and are designed to address critical knowledge gaps in the molecular heterogeneity of pediatric cancers and accelerate biomarker discovery, target identification, and translational research. To date, the RMC has completed its second year of data generation and submission to the CCDI data ecosystem, with WGS and RNA-seq data from ∼700 participants publicly released through the CCDI Hub, creating a growing community resource to advance pediatric cancer research. Preliminary multi-modal data integration using RNA-seq and whole slide images during the NCI Office of Data Sharing-hosted Childhood Cancer Data Jamboree in September 2025 demonstrated the added value of a comprehensive molecular picture to differentiate tumor subtypes and predict the presence of key pathogenic variants in CNS tumors with machine learning approaches. Similarly, integration of multimodal data in pediatric soft tissue sarcoma (STS) revealed four distinct immune phenotypes with the immune-high cluster exhibiting strong enrichment of inflammatory and immunosuppressive signatures, elevated expression of immune checkpoint genes, and increased tertiary lymphoid structure-associated signals and the immune-low cluster showing minimal immune activity and enrichment of metabolic pathways, supporting biologically meaningful stratification of pediatric STS for immunotherapy.