Sept - Soutenance de thèse de Floriane Poignant

Abstract

In radiation therapy, high-Z nanoparticles such as gold nanoparticles (GNPs) have shown particularly promising radiosensitizing properties. At an early stage, an increase in dose deposition and free radicals production throughout the tumour (photoelectric effect) and at sub-cellular scale (Auger cascade) might be responsible for part of the effect for low-energy X-rays. In this Ph.D work, we propose to study these early mechanisms with simulation tools, in order to better quantify them and better understand their impact on cell survival. We first finalised and validated Monte Carlo (MC) models, developed to track electrons down to low energy both in water (meV) and gold (eV). The comparison of theoretical predictions with available experimental data in the literature for gold provided good results, both in terms of secondary electron production and energy loss. This code allowed us to quantify the energy deposited in nanotargets located near the GNP, which is correlated with the probability to generate damages. This study required important optimisations in order to achieve reasonable computing time. We showed a significant increase of the probability of having an energy deposition in the nanotarget larger than a threshold, within 200 nm around the GNP, suggesting that GNPs may be particularly efficient at destroying biological nanotargets in its vicinity. The MC simulation was then used to quantify some chemical effects. At the macroscale, we quantified the increase of free radicals production for a concentration of GNPs. We also compared the radial distribution of chemical species following the ionisation of either a gold nanoparticle or a water nanoparticle. We showed that following an ionization, the average number of chemical species produced is higher for gold compared to water. However, in the vicinity of the nanoparticle, the number of chemical species was not necessarily higher for gold compared to water. This suggests that the effect of GNPs in its vicinity mostly comes from the increase of the probability of having an ionisation. We also studied several scenarios to explain the unexpectedly high experimental increase of the production of fluorescent molecules during the irradiation of a colloidal solution of GNPs and coumarin. Our study suggests that a plausible scenario to explain experimental measurements would be that GNPs interfere with an intermediate molecule, produced following the reaction between a coumarine molecule and a hydroxyl radical. During the last step of this Ph.D work, we injected our MC results in the biophysical model NanOx, originally developed at IPNL to calculate the biological dose in hadrontherapy, to predict cell survival in presence of GNPs. In addition, we implemented the Local Effect Model (LEM), currently the main biophysical model implemented for GNP-enhanced radiation therapy, to compare the NanOx and the LEM predictions with each other. In order to estimate cell survival with the LEM, we used various dosimetric approaches that were proposed in the literature. For a simple system where GNPs were homogeneously distributed in the cell, we showed that the LEM had different outcomes with regard to cell survival, depending on the dosimetric approach. In addition, we obtained an increase of cell death with the biophysical model NanOx that was purely due to the increase of the macroscopic dose. We did not obtain an increased biological effectiveness due to Auger electrons, which comes in contradiction with the LEM predictions. This study suggests that the current biophysical models available to predict the radiosensitizing effect of GNPs must be improved to be predictive. This may be done, for instance, by accounting for potential biological mechanisms evidenced by experimental works.