Mechanistic Modelling Of Radiobiological Effects On DNA Scale And Clinical Application

Proton radiotherapy is widely accepted as a cancer treatment modality due to its distinct advantage involving the Bragg Peak.A constant value for the relative biolog-ical effectiveness(RBE)of 1.1 is assumed in the clinical planning,despite of growing in-vivo evidence that RBE is a variant depending on many factors such as the physi-cal dose and biological endpoints.A number of phenomenological models scaled from the linear energy transfer(LET)have been developed to encapsulate variable RBEs.These models are based on linear quadratic equations,and their fitting parameters are derived from cell survival fraction(SF)data which tend to suffer from significant noise.The models,by definition,are not designed to reveal clue on underlying mechanisms of variable RBE.This doctoral thesis aims to bridge the gap between the physical dose and biological outcomes from the DNA molecular level to the organ level based on three publications.In achieving this aim,we perform four specific research tasks:(1)To simulate the radiation-induced DNA damage pattern for G0/G1 cell phase under normoxia;(2)To build hypothesized repair models in a step-by-step manner,(3)To derive RBE models with different biological endpoints,and(4)To implement the mod-els into treatment planning system by Monte Carlo simulated results.In this thesis,a comprehensive introduction of radiobiological effects is presented from the physical and biological aspects respectively.The mechanisms that lead to DNA damage and repair are simulated via Geant4-DNA and DaMaRiS tools,with parameters fitting to experimental results taken from the literature.The performances of the DNA dam-age/repair modeling are benchmarked against 37 cell lines from literature.Prediction of DNA damage and repair behaviours is correlated to conventional units,such as dose and LET,and RBE equations with four distinct biological-relevant end points,includ-ing initial DSBs,unrepaired DSBs,misrepaired DSBs and unrepair-misrepair combined DSBs,are deduced.These new models for RBE predictions are further compared with 365 experimental RBE values as well as three phenomenological RBE models in liter-ature using the dose and LET generated by a SOBP in a water phantom,and then two reasonable RBE models are applied to a clinical nasopharynx case.The results show that the damage data and repair models agree well with both experimental as well as simulated data reported in the literature to demonstrate the robustness of the models.The predicted initial DSB yields from damage model increase linearly with the increase of LET values,the misrepaired DSB fraction calculated from the resection-dependent repair model has a polynomial relation with LETs,and the unrepaired DSB fraction remains a constant of 5.8%along LET values.The water phantom results are clini-cally important in that RBE value predicted solely from misrepair DSB is higher by as much as two orders of magnitude than those from unrepaired DSB or the combination of unrepair and misrepair DSBs.This work also fills the gap between the physical dose and biological outcomes in terms of multi-scale DNA molecular and organ level end points.The work shows that the mechanistic models are indeed more useful in predict-ing biological outcomes,and,therefore,follow-up patient studies based on such models can be performed to validate this powerful and multi-centric tool in the ultimate role of optimizing the biological dose in treatment planning.