Identification of critical Monte Carlo simulation parameters in nanoparticles radiosensitization

International audienceThe use of nano-objects as radiosensitizers is rapidly increasing and has become a major subject of interest as a new strategy to fight cancer. This concept has already been proven in vitro and in vivo by a number of authors (Butterworth et al., 2012) but only a few nanoparticles (NPs) are undergoing clinical trials. Even though a lot of preclinical studies are con-ducted to investigate the effects of nano-objects combined with X-rays on cells/tissues/tumor, there is still a serious lack of knowledge about their physical and chemical mechanisms.A common way to investigate their properties is to use Monte-Carlo simulation methods with particle transport codes such as Geant4 (Agostinelli et al., 2003). These simulations are time consuming, strongly code or model dependent and do not com-pare well between each other. It has already been noticed that the lack of standardization in preclinical studies of nano-radiosensitizers could partially ex-plain the low number of translation to clinical appli-cations (Retif et al., 2015). That seems to become problematic in the field of NPs (Kodiha et al., 2015).In this study, a sensitivity analysis is performed to estimate the impact of eight simulation parameters on the variability of two numerical responses: the dose enhancement and the dose diffusion in seven spatial regions of interest. The physical modeling and Monte Carlo simulations are supported by the Geant4-GATE environment. The simulated nano-object is a single 100 nm gold nanoparticle. A Box and Behnken design of numerical experiments was carried out to estimate the statistical relevance of the simulations factors.Finally three simulation parameters: the Auger ef-fect, the medium type and the fluence level, have been identified as critical factors and therefore have to be chosen carefully and to be kept constant be-tween interlaboratory comparisons of numerical simulations. Another factor: the cutoff energy may have a significant impact on the diffusion response. However, whatever the studied response, four simu-lation parameters have negligible effects: the work-ing volume, the spatial resolution and cutoff, and the computer. Those results bring new contributions to improve the robustness of numerical simulations of nanoparticles activated by X-ray and proposes direc-tions for standardization in this scientific area