Geant4 modelling of DNA strand breaks from diffusing alpha-emitters radiation therapy

Abstract

α-therapies have shown promise for the treatment of cancer. One such approach is diffusing alpha-emitters radiation therapy (DaRT), where seeds containing ²²⁴Ra are inserted into tumours. Monte Carlo track structure simulations (Geant4 and Geant4-DNA) were used to simulate the DNA damage due to DaRT.

Mono-energetic simulations of the α-particles in DaRT showed the high LET of the α-particles increases DNA damage complexity compared to low LET radiotherapy. The relative biological effectiveness (RBE) for double strand break (DSB) induction is the range 2–8. The high LET also reduces the proportion of indirect damage, with a maximum direct to indirect strand break ratio of 0.46 ± 0.01 for α-particles compared to 0.19 ± 0.01 for ⁶⁰Co.

Simulations of all four phases of the cell cycle showed that the relative sensitivity in terms of DSB induction from 9 MeV α-particles is G2>M>S>G1. Cancer cell lines showed greater yields of DNA damage compared to normal cells for all phases. Normal human fibroblast cells in a resting state were calculated to have a DSB yield of 43.8 ± 1.0 Gy^-1cell^-1. In contrast, the yield of DSBs in asynchronous populations of cancer cells were calculated to be 96.3 ± 7.0 Gy^-1cell^-1 and 100.2 ± 5.8 Gy^-1cell^-1 for U2OS and HeLa cell lines.

DaRT seed simulations showed that the particle distribution and DNA damage induced vary as a function of distance from the seed resulting in a variable RBE between 1–7. The required DaRT seed spacing was estimated to be 5.5 mm and 6.5 mm using RBE for DSB and complex DSB induction. Simulations were fitted to experimental measurements to assess how additional migration of ²²⁰Rn in the blood flow affects treatment planning and dose calculations. The additional migration causes an increase in RBE weighted dose distal to the DaRT seed. The area enclosed by the 20 Gy and 1 Gy RBE weighted isodose lines increases by 23.9 ± 0.3% and 30.3 ± 
0.1%.

A fast Monte Carlo method is presented to reduce the computational burden of such simulations. Simulation time for the DNA damage from a DaRT seed was reduced from approximately 20 weeks of CPU hours to less than 0.25 CPU hours. Comparing the results of the fast method and Geant4 simulation for DaRT the mean percentage difference was -0.4%, -1.7% and 0.2%, and standard deviation 2.8%, 4.8% and 1.8% for dose, number of strand breaks and number of DSBs. Thus, for all three variables the mean percentage difference is within one standard deviation of zero indicating that the method can reproduce the full simulation results.

DaRT induced DNA damage has been shown to be complex and therefore likely to be lethal. However, DSB induction efficiency changes as a function of distance from the DaRT seed. Migration parameters in the model influence DSB induction and the calculated dose distribution. Simulations have shown how these effects can be used to inform and optimise treatment planning.

Date of Award	30 Sept 2025
Original language	English
Awarding Institution	University of Bristol
Supervisor	Lana Beck (Supervisor) & Jaap J Velthuis (Supervisor)