Project Summary Conventional wisdom holds that radiation therapy is a physically targeted anti-cancer modality and that its cancer targets are genetically and biologically uniform. However, the stromal composition of tumors is complex and variable giving rise to extrinsic variability of the cancer target. Moreover, recent studies using single cell genomics show intrinsic heterogeneity even within the tumor cells per se. The broad goal of our Harvard/UCSF ROBIN initiative is to test the hypothesis that intra-tumoral variability underlies resistance to - and relapse from – radiotherapy. Towards this goal, our Center has chosen to focus on pediatric cancers of neuro-ectodermal origin. Pediatric tumors have a low mutational burden relative to common adult cancers and thus a cleaner genetic surround for the “low N/high content” Molecular Characterization Trials specified by the ROBIN RFA. Against this backdrop, this project focuses upon diffuse midline glioma (DMG) of children. The majority of DMGs initially respond to radiation, but all progress, and none are cured. In preliminary studies, we have used single cell genomics to show that the malignant cells within DMG are developmentally heterogeneous. Our testable hypothesis is that DMG intratumoral heterogeneity transcends developmental markers to include differential expression/utilization of common DNA repair pathways. This hypothesis makes predictions that will be assessed by drawing upon paired samples of pre-and post-radiotherapy tumor tissue from children treated prospectively with a uniform radiotherapy regimen and profiled in our Molecular Characterization Trial (MCT). We have three specific aims: Aim 1 is to test the prediction that intratumoral heterogeneity is reflected at levels above and beyond tumor cell-specific developmental markers noted our preliminary studies; Aim 2 is to test the prediction that radiotherapy reduces DMG intratumoral heterogeneity via selection of replication-competent, radio-resistant stem-like cancer cells; Aim 3 is to test the prediction that heterogeneous radiation responses within tumor cells underlie patient heterogeneity in radiation-associated toxicities, neurocognitive effects and quality of life. The study plan incorporates contemporary methods in cancer genomics, epigenomics, chromatin biology and DNA enzymology. We will draw upon our Clinical Artificial Intelligence and Imaging Core to develop non-invasive methods to track intra-tumoral heterogeneity in these (surgically challenging) pediatric tumors. With our Molecular Data Science and Advanced Dosimetry Core, we will develop computational modeling of tumor cell evolution and treatment response that will be critical to understanding selection for radioresistant subclones. The co-leaders of this Project have complementary expertise to enable the study plan. Daphne Haas-Kogan, M.D. is a pediatric radiation oncologist who treats patients with DMG and holds leadership positions in two key consortia (COG ...