Monte Carlo simulations of nanodosimetry and radiolytic species production for monoenergetic proton and electron beams: Benchmarking of GEANT4‐DNA and LPCHEM codes

摘要

Abstract Purpose In hadrontherapy, biophysical models can be used to predict the biological effect received by cancerous tissues and organs at risk. The input data of these models generally consist of information on nano/micro dosimetric quantities and, concerning some models, reactive species produced in water radiolysis. In order to fully account for the radiation stochastic effects, these input data have to be provided by Monte Carlo track structure (MCTS) codes allowing to estimate physical, physico‐chemical, and chemical effects of radiation at the molecular scale. The objective of this study is to benchmark two MCTS codes, Geant4‐DNA and LPCHEM, that are useful codes for estimating the biological effects of ions during radiation therapy treatments. Material and methods In this study we considered the simulation of specific energy spectra for monoenergetic proton beams (10 MeV) as well as radiolysis species production for both electron (1 MeV) and proton (10 MeV) beams with Geant4‐DNA and LPCHEM codes. Options 2, 4, and 6 of the Geant4‐DNA physics lists have been benchmarked against LPCHEM. We compared probability distributions of energy transfer points in cylindrical nanometric targets (10?nm) positioned in a liquid water box. Then, radiochemical species (· OH, eaq?${rm{e}}_{{rm{aq}}}^ - $, H3O+,H2O2${{rm{H}}_3}{{rm{O}}^ + },{rm{;}}{{rm{H}}_2}{{rm{O}}_2}$, H2, and OH?)${rm{O}}{{rm{H}}^ - }){rm{;}}$yields simulated between 10?12 and 10?6 s after irradiation are compared. Results Overall, the specific energy spectra and the chemical yields obtained by the two codes are in good agreement considering the uncertainties on experimental data used to calibrate the parameters of the MCTS codes. For 10 MeV proton beams, ionization and excitation processes are the major contributors to the specific energy deposition (larger than 90) while attachment, solvation, and vibration processes are minor contributors. LPCHEM simulates tracks with slightly more concentrated energy depositions than Geant4‐DNA which translates into slightly faster recombination than Geant4‐DNA. Relative deviations (CEV) with respect to the average of evolution rates of the radical yields between 10?12 and 10?6 s remain below 10. When comparing execution times between the codes, we showed that LPCHEM is faster than Geant4‐DNA by a factor of about four for 1000 primary particles in all simulation stages (physical, physico‐chemical, and chemical). In multi‐thread mode (four threads), Geant4‐DNA computing times are reduced but remain slower than LPCHEM by ～20 up to ～50. Conclusions For the first time, the entire physical, physico‐chemical, and chemical models of two track structure Monte Carlo codes have been benchmarked along with an extensive analysis on the effects on the water radiolysis simulation. This study opens up new perspectives in using specific energy distributions and radiolytic species yields from monoenergetic ions in biophysical models integrated to Monte Carlo software.