PO-197 Patient-derived xenograft (PDX) models of glioblastoma: from basic research to preclinical studies

Abstract

IntroductionAnimal models are essential tools for basic research and preclinical therapeutic interventions. Although numerous clinical cancer trials are being conducted, many fail due to inappropriate selection of compounds at the preclinical stage. Therefore better preclinical models are crucial for predicting successful clinical impact. Orthotropic PDX models are of particular importance for brain cancers, as they allow to better recapitulate the brain tumour environment and the blood brain barrier.Material and methodsGlioblastoma PDX models were based on 3D organotypic spheroids, derived from mechanically minced patient material. Spheroids were implanted in the brain of immunodeficient mice and further propagated by serial intracranial transplantations. For detailed molecular characterisation each PDX was compared to its original patient tumour at the genetic, epigenetic and transcriptomic levels and intra-tumoral heterogeneity was addressed at the single cell level. We furthered performed proof-of-concept preclinical studies interfering with angiogenesis and autophagy.Results and discussionsOur glioblastoma PDX models starting with viable patient-derived spheroids has a high tumour take rate and a reproducible phenotype and tumour development time. We observed three distinct histological tumour phenotypes: a highly ‘invasive’, a highly ‘angiogenic’ and an ‘intermediate’ phenotype which combines invasion and vascular abnormalities. Typical glioblastoma characteristics such as pseudopalisading necrosis, invasion or microvascular proliferation we maintained. PDXs retained the genetic and epigenetic profiles of patient tumours through several generations. Transcriptomic profiles of PDXs were similar to patient biopsies and correlated better with TCGA glioblastoma samples than conventional glioma cell lines. In vivo pharmacological inhibition of autophagy significantly increased survival of PDXs and combination treatment with bevacizumab showed a synergistic effect.ConclusionHere we show that glioblastoma PDXs represent a reliable and clinically-relevant animal model. The model can be applied for analyses at different molecular levels. Importantly, the PDXs can be applied for accurate reproducible pre-clinical trials, including personalised medicine-based treatments. The use of this model should lead to a more realistic evaluation of the efficacy of novel drugs, thereby increasing the success of clinical studies.