Modelling DNA damage and repair in radiation therapy

Abstract

Radiation therapy is a vital tool in the treatment of cancer, with the primary aim of maximising tumour dose while minimising healthy tissue exposure. Historically, most patients received X-ray therapy, but recent years have seen a movement towards ion therapy which has superior tumour targeting and reduced healthy tissue exposure. Despite this enhanced spatial conformality of ions compared to X-rays, their use remains limited due to the cost of the large accelerators required for treatment delivery. There is however growing evidence that ion therapy is also more biologically effective than X-ray therapy, requiring a lower dose to achieve the same treatment outcome, which could further its benefits. Through increased understanding of how ions interact within cells, these additional benefits of ion therapy can be better quantified, to provide robust evidence for more effective clinical use of ion therapy.

Radiation induces cell death by interacting with and causing damage to the DNA. The cell’s ability to repair this damage determines the overall response to irradiation. Computational models that simulate DNA damage and repair offer predictive tools to study radiation interactions in biological systems, enabling the biological effectiveness of different radiation treatments, and their underlying mechanisms to be explored. In this work, the TOPAS-nBio toolkit was used to simulate radiation interactions and DNA damage, alongside the Medras model which simulated repair and biological response to this damage. This work aimed to answer three main questions. The first investigated if changing the dose-rate for proton delivery increased the chance of successive protons interacting with each other, in a way that would impact biological response compared to conventional dose-rates. The second explored how differences in DNA damage model design impacted on predicted DNA damage. And finally, the third question investigated if these differences in damage model design impacted on predicted biological response following repair.

It was found that at clinically relevant doses, dose-rates many orders of magnitude higher than conventionally used are unlikely to result in any additional proton-proton interactions that could contribute to differential treatment outcomes. Investigations of DNA damage model design found that by appropriate parameter fitting, all models could predict damage yields in line with the experimental data, across a range of radiation qualities. This suggests that damage model design does not fundamentally impact predicted damage, with simpler, more efficient models having a similar predictive power for both initial damage and later endpoints. Together, this work has shown how computational models of radiation response are valuable predictive tools, while also being mechanistically informative. Additionally, areas of model refinement and experimental validation have been identified that would help further advance these models for future applications.

Date of Award	Jul 2024
Original language	English
Awarding Institution	Queen's University Belfast
Sponsors	Northern Ireland Department for the Economy
Supervisor	Stephen McMahon (Supervisor) & Kevin Prise (Supervisor)