Simplified Estimation of X-Ray Induced DNA Damage from Monte Carlo Track-Structure Data
Abstract
Purpose
To develop a fast and scalable computational framework for estimating DNA damage induced by X-ray irradiation while retaining track-structure fidelity and reducing computational cost.
Methods
We constructed an energy-specific database of DNA damage events using Geant4-DNA and TOPAS-nBio. This database served as the foundation for the computational pipeline to efficiently estimate the radiation-induced DNA damage outcomes without rerunning computationally intensive full track-structure simulations. We employed a multiscale simulation approach in which the macroscale simulation modeled the X-ray beam over a water phantom, while the microscale simulation modeled DNA damage within a detailed nucleus geometry. Photon energies between 60 keV and 10 MeV were simulated to cover all clinical and preclinical cases. For each possible scenario of energy deposition, we recorded the amount of deposited energy and the number of different types of DNA damage. In each simulation, the cell received approximately 1 cGy of absorbed dose.
Results
Across all energies, the average number of double-strand breaks per unit dose (DSB/Gy) was consistent with the commonly reported values of approximately 40 DSB/Gy. The widespread distribution of DSB/Gy values across energies reflects the stochasticity in radiation and DNA interactions, and the outliers represent rare but biologically possible events where a single track deposits energy in either a highly sparse or a highly clustered manner. In addition to total DNA damage counts, the pipeline classified each break by damage complexity. The output shows that more than 50% of the DSB exhibited low complexity, with only a small fraction associated with highly clustered damage.
Conclusion
While a full track-structure simulation for a single cell could take several weeks, the pipeline allows simulation of 6,000 cells within only a few hours. Unlike conventional approaches limited to a single cell, this work can be extended to tissue-scale modeling.