ID #1043 BioBombe-Guided Pathway Analysis Identifies Candidate Therapeutics in Pediatric High-Grade Glioma

Background

Pediatric high-grade gliomas (pHGG) are highly aggressive childhood cancers and remain largely refractory to current therapies. This is due to oncogenic drivers converging at the pathway level , rather than single mutations. To address this, we applied BioBombe, an in-silico framework that integrates genome-wide CRISPR vulnerability screens (DepMap Project Achilles), drug viability screens (DepMap PRISM assay), bulk RNA-seq, and pathway analysis, to predict and interpret drug sensitivity at the pathway level.

Methods

We previously trained a BioBombe model on Project Achilles CRISPR screen data (Curd and Way 2025 )[1]. We applied this model to six pHGG cell lines and 14 primary pHGG tumor samples obtained from Children’s Hospital Colorado. For each sample, genesets reflecting key oncogenic pathways were mapped to associated drugs. Drug frequencies were aggregated across samples to identify recurrent individual drug hits and enriched drug classes (e.g., DNA damage, mitotic inhibitors, TKIs). Independent analysis of MAPK pathway genes was used to evaluate whether BioBombe-identified drug classes were aligned.

Results

BioBombe identified highly consistent enrichment of specific drug classes. The most prominent classes included DNA repair/synthesis inhibitors, mitotic inhibitors, and tyrosine kinase inhibitors (TKIs). BioBombe identified individual drugs targeting DNA synthesis, mitosis, RTK signaling, and angiogenesis. Among BBB-penetrant drugs, cladribine was recurrent across brain tumor samples within the DNA repair/synthesis inhibitor class, while axitinib consistently mapped to the TKI/angiogenesis drug class. Further independent analysis of MAPK pathway genes revealed significant and sustained dysregulation of upstream RTKs and ligands. Preliminary functional validation using MTS viability assays demonstrated consistent reductions in metabolic viability following treatment with both agents across tested model

Discussion

This study demonstrates that BioBombe can identify drug candidates in pediatric high-grade glioma by integrating transcriptomic pathway activity with drug–gene association data. Rather than prioritizing isolated targets, BioBombe highlights drugs that intersect with shared oncogenic programs across tumors. Preliminary functional validation supports BioBombe-derived predictions.

1. (Curd and Way 2025 )