The trap of 2D in artificial models of tumours: The case for 3D in-silico simulations

Artificial modelling of tumours can provide insights in cancer biology and offer a powerful complement to laboratory research. A common approach is to simulate tumour growth in a two-dimensional environment and to then generalize results to a three-dimensional one. Literature suggests this strategy fails to adequately capture the underlying biology and may provide misleading results. To establish whether 2D models may form a viable alternative to 3D ones, we developed a model comprising cancer cell growth and proliferation and soluble diffusion to replicate experiments previously performed in a laboratory. We made use of established parametrization techniques to configure our simulations and novel error estimation strategies to evaluate them. Our results suggest that the same simulation in 2D and 3D yields significantly different results. Further, that the cause of this discrepancy lies in the spatial geometry of 2D simulations which does not allow for the formation of hypoxic regions in the tumour mass. We conclude with a recommendation that due to the limitations of 2D simulations, and the negligible difference in cost between the two approaches, 3D simulations should be employed over 2D ones.