ID #886 Landscape of genetic ancestry-associated active drugs unique to or shared by White and Hispanic children with group 3 medulloblastomas

Abstract

Pediatric brain tumors remain the leading cause of cancer-related mortality in children, and disparities in clinical outcomes have been observed across racial and ethnic groups. While socioeconomic and socio-environmental factors contribute to these disparities, the extent to which ancestry-associated tumor biology influences therapeutic response remains poorly understood. Progress in this area has been limited by the availability of biologically relevant model systems and the lack of functional studies comparing drug responses across ancestry-associated tumor models. From a collection of > 160 PDOX models, we inferred genetic ancestry for 112 models using RNA-seq–based analysis and SNP genotyping with the Infinium® Global Diversity Array, with GDA providing higher-resolution classification and enabling detection of mixed-ancestry tumors beyond self-reported racial/ethnic information.

Here, we applied high-throughput functional drug screening to patient-derived xenograft (PDOX) models of pediatric brain tumors, with a focused analysis on group 3 medulloblastoma, the most aggressive molecular subtype. Using xenograft-derived tumor cells cultured under both fetal bovine serum medium (FBS) and cancer stem cell medium (CSC), we performed drug screening across a library of approximately 8,000 FDA-approved and investigational compounds in six White- and six Hispanic-derived group 3 medulloblastoma models. Drug response was quantified using area under the curve (AUC)–based thresholds (0.7–1.0) after 7 days of treatment to define active compounds.

This analysis revealed distinct ancestry-associated drug response patterns. Compounds selectively active in Hispanic-derived models but not in White-derived models reached 27 clustered predominantly around targets and pathways related to CDK-mediated cell cycle regulation (Dinaciclib), Na+/K+ ATPase–associated calcium signaling, and mTOR-linked reactive oxygen species and apoptosis signaling. In contrast, compounds preferentially active in White-derived models reached 56 were enriched for targets involved in topoisomerase-mediated DNA damage responses (Irinotecan) and microtubule-and HSP-associated cytoskeletal signaling pathways. These differences were observed at the level of pathway and target enrichment rather than individual drug effects, suggesting systematic differences in therapeutic vulnerabilities between ancestry-associated tumor groups.

Together, our study uncovers reproducible genetic ancestry-based drug response profile between white and Hispanic G3 MBs, provides a biologic rational for mechanistic investigation of cancer disparities and supports the development of ancestry-informed precision therapeutic strategies for medulloblastoma.