What have we learnt from previous phase II trials to help in the management of childhood brain tumours?

Contrary to major advances in cure rates observed for almost all childhood cancers, progress in reducing brain tumour survival rates remains very limited. Although new drug development in oncology is founded on principles outlined in the organised methodology of phase I, II, and III trials, based on rigorous study design using standardised criteria, this approach has been applied very slowly in the field of neuro-oncology. There are multiple explanations for the paucity of well-conducted prospective clinical trials, such as the rarity and the heterogeneity of these tumours, and the reluctance of some investigators to enrol their patients in constraining trials. Data from the past two decades shows that several methodological problems preclude the drawing of any definite conclusions for the majority of drugs assessed. Among them, the necessity of a central neuropathological and neuroradiological review has been highlighted in, at least, two multicentric studies. Changes in histological diagnosis and grade have been reported in a proportion as high as 20%, and changes in response rate in 14% of the cases. This review of phase II trials for brain tumours reveals a wide array of sometimes arbitrary response definitions, that is if response is defined at all, and most series have enrolled small numbers of patients. We report on the different problems encountered in childhood brain tumours in these phase II trials, and their impact on phase III trials.