Thomas R.W. Oliver1,2 and Sam Behjati2,3,4 1Department of Histopathology and Cytology, Cambridge University Hospitals NHS Foundation Trust, Cambridge, Cambridgeshire CB2 0QQ, United Kingdom 2Wellcome Sanger Institute, Hinxton, Cambridgeshire CB10 1RQ, United Kingdom 3Department of Paediatrics, University of Cambridge, Cambridge, Cambridgeshire CB2 0QQ, United Kingdom 4Department of Paediatric Haematology and Oncology, Cambridge University Hospitals NHS Foundation Trust, Cambridge, Cambridgeshire CB2 0QQ, United Kingdom Correspondence: to3sanger.ac.uk; sb31sanger.ac.uk

Most childhood cancers possess distinct clinicopathological profiles from those seen in adulthood, reflecting their divergent mechanisms of carcinogenesis. Rather than depending on the decades-long, stepwise accumulation of changes within a mature cell that defines adult carcinomas, many pediatric malignancies emerge rapidly as the consequence of random errors during development. These errors—whether they be genetic, epigenetic, or microenvironmental—characteristically block maturation, resulting in phenotypically primitive neoplasms. Only an event that falls within a narrow set of spatiotemporal parameters will forge a malignant clone; if it occurs too soon then the event might be lethal, or negatively selected against, while if it is too late or in an incorrectly primed precursor cell then the necessary intracellular conditions for transformation will not be met. The precise characterization of these changes, through the study of normal tissues and tumors from patients and model systems, will be essential if we are to develop new strategies to diagnose, treat, and perhaps even prevent childhood cancer.