Presenter(s)
N. Joshi1, D. T. Bergman2, S. Nellore1, P. Chen3, R. H. Nanda Jr4, K. Yamoah5, S. Stapleton6, J. Durkin1, A. Durmaz7, S. A. Eschrich8, J. F. Torres-Roca5, and J. G. Scott9; 1Cleveland Clinic Lerner College of Medicine, Cleveland, OH, 2Geisel School of Medicine at Dartmouth, Hanover, NH, 3Case Western University, Cleveland, OH, 4Morton Plant Hospital, Clearwater, FL, 5Department of Radiation Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, 6Johns Hopkins All Children's Hospital, St. Petersburg, FL, 7Cleveland Clinic, Cleveland, OH, 8Department of Biostatistics and Bioinformatics, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, 9Department of Radiation Oncology, Taussig Cancer Institute, Cleveland Clinic Foundation, Cleveland, OH
Biology-guided radiosensitivity in pediatric brain tumors provides insight into intrinsic radiation biology and enables personalized radiotherapy frameworks.
Purpose/Objective(s):
Radiotherapy (RT) is a central component of treatment for pediatric brain tumors but is prescribed using histology-driven doses without accounting for underlying biologic heterogeneity. The radiosensitivity index (RSI) and genomic-adjusted radiation dose (GARD) have been validated in adult cancers to quantify intrinsic radiosensitivity and biological radiation effect, but their role in pediatric CNS tumors is not well defined. We therefore evaluated RSI-driven tumor biology and assessed whether GARD predicts RT response in pediatric brain tumors to inform biologically personalized radiotherapy in this population.
Materials/Methods:
We performed a retrospective analysis of samples from the Children’s Brain Tumor Network (CBTN). RSI was generated from a 10-gene expression signature and integrated into GARD through the linear-quadratic model. Top-quartile RSI samples was compared to bottom-quartile RSI samples utilizing differential gene expression, pathway analyses and cellular deconvolution. Cox proportional hazards models were used applying GARD as a continuous variable with overall survival and local recurrence as endpoints.Results:
The cohort included 69 high-grade glioma, 64 ependymoma, and 74 medulloblastoma patients. Median age was 8.9 years old. Medulloblastoma was predicted to be most radioresistant (RSI = 0.75), followed by high-grade glioma (RSI = 0.54), and ependymoma (RSI = 0.49). While RT dose was uniform across tumor types (Median: 54 Gy, SD: 2.5 Gy), GARD varied widely (Median: 16.2, SD: 9.7), Radiosensitive tumors were enriched for immune and inflammatory signaling programs, including TNF-a, IFN-?, and related inflammatory pathways (all p < 0.001), relative to radioresistant tumors. Deconvolution analyses further demonstrated increased microglial and endothelial cell fractions in radiosensitive tumors (p < 0.001), consistent with a distinct immune–microenvironmental phenotype. GARD was independently associated with improved overall survival (HR = 0.96/unit, p = 0.04) and local recurrence (HR = 0.95/unit, p < 0.001), corresponding to a 35% reduction in mortality risk and 50% reduction in recurrence risk per 10-unit increase.Conclusion:
Radiosensitivity in pediatric brain tumors demonstrates strong biologic heterogeneity characterized by immune and microenvironmental signaling differences. Additionally, GARD was significantly associated with overall survival and local recurrence, supporting its utility as a biologically informed predictor of radiation effect. These findings suggest RSI- and GARD-based approaches may facilitate biologically personalized radiotherapy dose strategies in pediatric central nervous system tumors.