STEM-08. Bioprocessing of viable surgical pediatric brain tumor specimens for genome-guided personalized drug testing in conjunction with the Human Cancer Model Initiative

Abstract Novel treatment approaches are urgently needed for pediatric central nervous system (CNS) tumors as they are the leading cause of cancer-related deaths in children. A lack of research materials impedes progress towards developing such therapies. Although various brain cancer biobanks exist, these rarely store viable patient-derived tissue and cells for live cell analyses, including cell fate assays and testing for drug sensitivities. A biorepository of viable cell dissociates and tissue representing CNS tumor entities could mitigate these limitations. From over 130 samples collected, we generated a biorepository of 39 distinct entities of pediatric CNS tumors in our Stanford neuro-bioprocessing cohort. Based on the Central Brain Tumor Registry of the United States, the proportion of subtypes represented in our cohort match that of the general population with exceptions. We established a standardized bioprocessing pipeline that can be used as a template for tissue collection and model development in the broader disease context and for multiple downstream applications. Pediatric brain tumor model development in vitro poses challenges related to genetic drift impairing fidelity to the parent tumor. We will present optimized growing conditions for several pediatric low grade glioma models alongside their validation by immunophenotyping and molecular analysis. Comparing RNA expression profile of an individual patient’s pilocytic astrocytoma with a compendium of 12,747 brain and non-brain tumor cases revealed outliers. These were orthogonally tested in patient-matched cancer models using pharmacological inhibitors, validating them as potential novel drug targets. These studies illustrate that biobanking of viable patient-­derived material enables developing personalized therapies with the goal of improving patient outcomes. Conclusively, patient-derived cancer models facilitate identifying and validating drug vulnerability, creating an opportunity for rapid translation of personalized treatment for patients with CNS tumors.