EX-VIVO 3-DIMENSIONAL MODELS OF POST-SURGICAL RESIDUAL DISEASE IN HUMAN GLIOBLASTOMA

Abstract Translation of novel therapies for glioblastoma from laboratory to clinical trials has relied heavily on cells derived from within patient tumours. However, it is clear from over 40 years of limited improvement in survival rates, that the traditional paradigm of 2-dimensional in-vitro studies using these cells followed by implantation into mouse models poorly predicts the treatment response of glioblastoma cells left behind after surgery in patients. Additionally, recent increases in the number of new therapies requiring pre-clinical evaluation emphasise a need to apply disease models that faithfully and efficiently reproduce biological features of post-surgical residual glioblastoma in patients. Dependent on the anatomical considerations of individual cases, surgical resection occasionally provides an en-bloc specimen (e.g. partial lobectomy), which includes portions of infiltrated brain that would not otherwise be resected. Such specimens provide crucial material to model and understand residual (typically left behind) disease in glioblastoma. Using such specimens, we have developed a pragmatic methodology to liberate glioblastoma stem cells (GSC) from the tumour core and adjacent infiltrated brain to model resected and residual disease, respectively. Imperative considerations include careful patient selection and intraoperative clinical-academic collaboration, with immediate ex-vivo hemi-section and processing of suitable specimens. Initial biological characterisation of our resected and residual models reveals numerous divergent properties including morphological differences and elevated stem cell marker expression in residual models (p=0.0021). Importantly, although GSC within recently derived 2-dimensional residual disease models are frequently too migratory to reliably form discrete colonies, we are able to perform robust clonogenic survival studies using the same GSC within a novel customised 3-dimensional scaffold-based architecture (3.4-fold greater clonogenic potential, p=0.0010). In conclusion, we believe that our models provide a biologically relevant and cost-effective tool to investigate the nuances of residual glioblastoma and assess the potential of therapies to tackle disease normally left behind following surgery in patients.