A DNA damage calculation method for large-scale tumor models in boron neutron capture therapy

Objective, Boron neutron capture therapy (BNCT) is considered a binary, targeted radiotherapy modality that can effectively eradicate tumor cells via the 10B(n,α)7Li nuclear reaction. However, because conventional Monte Carlo approaches suffer from severe computational inefficiency in DNA damage calculations, most studies have been confined to single-cell models, which are inadequate for characterizing the spatial distribution of DNA damage in clinically relevant tumor tissues. To address this limitation, we propose a method that quantitatively evaluates DNA damage yields induced by α and 7Li particles in BNCT at an acceptable computational cost. Approach. We developed a fast DNA damage evaluation framework based on Geant4 and its extension Geant4-DNA. First, Geant4-DNA was used to obtain DNA damage yields induced by α and 7Li particles within a unit volume, from which an energy-dependent DNA damage yield database was constructed. The DNA damage data were then integrated into a customized Geant4 program to track particle transport, and the particle energy-step-length information was mapped to the corresponding DNA damage yields, enabling rapid damage estimation in centimeter-scale models. Main results. Using the proposed method, DNA damage yields for more than one million α and 7Li particles can be computed within tens of hours in a single-cell model. The results show that the 10B distribution associated with BPA yields higher DNA damage yields and higher damage probabilities than BSH. In a 1.44 cm × 1.44 cm × 1.8 cm tumor model, the cell-averaged DNA damage yield varies with deposited dose and fluctuates within a certain range, indicating non-uniform microscopic dose deposition governed jointly by neutron-spectrum evolution and the transport of reaction products. At the centimeter scale, BPA still produces higher overall damage than BSH, although the difference between them becomes less pronounced than in the single-cell case. Significance This study applies, for the first time, to a large-scale tumor model and estimates nanoscale DNA-damage yields in tumor cells induced by7Li and α particles within an acceptable computational cost. These findings will support further analysis of the spatial distribution of DNA damage during clinical treatment.