Modeling human carcinomas: physiologically relevant 3D models to improve anti-cancer drug development.

Anti-cancer drug development is inefficient, mostly due to lack of efficacy in human patients. The high fail rate is partly due to the lack of predictive models or the inadequate use of existing preclinical test systems. However, progress has been made and preclinical models were improved or newly developed, which all account for basic features of solid cancers, three-dimensionality and heterotypic cell interaction. Here we give an overview of available in vivo and in vitro models of cancer, which meet the criteria of being 3D and mirroring human tumor-stroma interactions. We only focus on drug response models without touching models for pharmacokinetic and dynamic, toxicity or delivery aspects.