Abstract ID: 68 Heterogeneous multiscale simulations of radiation therapy with gold nanoparticles

Abstract

Purpose An outstanding challenge for gold nanoparticle (GNP) dose-enhanced radiation therapy (GNPT) is accurate computation of energy deposition in microscopic volumes of interest within macroscopic tumour and normal tissues. This work presents calculations for GNPT within the Heterogeneous MultiScale (HetMS) model, a general Monte Carlo framework. Methods The HetMS approach involves combining distinct models of varying level of detail on different length scales within a single simulation. Using EGSnrc, GNPT HetMS models considering varying extra/intracellular GNP distributions are implemented. Groups of > 3000 cells containing discretely-modelled GNPs are embedded at different positions throughout a cylindrical phantom (2 cm diameter, 3 cm length) consisting of a gold-tissue mixture. Dose enhancement factors (DEFs; ratios of doses with and without gold present) are calculated for cell nucleus and cytoplasm compartments for various photon source energies, cell/nucleus sizes, and gold concentrations. HetMS simulations are validated via comparisons with simulations of > 1.5E10 discretely-modelled cells and GNPs throughout the cylindrical phantom. Results DEFs are highly sensitive to depth in phantom, GNP distribution and concentration, cell/nucleus size, and source energy. Considering an isolated cell (not at depth in phantom), nuclear DEFs change with cell/nucleus and GNP intracellular distribution, ranging from 1.6 (GNPs in an endosome, 5 mg of gold/g of tissue, 50 keV photons) to 18.7 (GNPs just outside the cell nucleus, 20 mg/g, 20 keV). Nuclear DEFs are consistently higher for GNPs clustered about the nuclear membrane compared to GNPs within endosomes in the cell cytoplasm. Nuclear DEFs decrease considerably with depth in the phantom, e.g., changes of up to 64% are observed over the first centimeter. HetMS simulation results are in agreement (within statistical uncertainties) with those of simulations of discrete cells and GNPs throughout the phantom, but are considerably more efficient with HetMS simulations > 1000 times faster. Conclusions This work demonstrates considerable variation in DEFs, due to effects such as decreasing fluence because of gold in the macroscopic phantom and local enhancement because of nanoparticle distribution within cells. Results underline the importance of quantifying DEFs across tumour volumes while considering intracellular GNP distribution. The HetMS approach enables accurate and relatively efficient simulations of GNPT.