The second topic was today’s panel discussion on the use of pre-clinical models. I was asked to chair this session because I am not a pre-clinical modellist; I am actually a clinical triallist who counts on my colleagues who do the pre-clinical testing to advise me which are the best new treatments to take forward. We have all recognised that there is not a very good track record of pre-clinical testing actually predicting clinical utility so we were bringing up the various issues related to that - what types of model systems, we heard some very good examples and some overviews. Actually I, in my opening comments, really advocated for there to be a central repository, much like we do for genomic data of patients’ cancer information, but to do the same for pre-clinical testing – what the testing was, what the system was, what the results were, and then be able to then in the future correlate that with what happens in the context of a clinical trial. So do advanced informatics and there are systems that have been established, database systems. There’s one I mentioned called Monarch which has been used for genetic disorders to compare the genotype and phenotype across species so that you can look at other species and models in those species in order to inform the human disease .No reason we can’t adapt a system like that for brain cancer or cancer in general.