Oxygen Related Factors in FLASH Radiotherapy

Abstract

To develop an oxygen depletion model for pulsed ultra-high dose rate radiation therapy. To verify the model against published experiments in-vitro and in-vivo. Finally, we estimate the impact of dosimetric variations on the quantification of the protective effect on tissue. A previously developed model of oxygenation response on the creation of clustered DNA-damage (Stewart PMB 2015, Van den Heuvel PLosOne 2014) is combined with a published mechanistic agnostic empirical model of oxygen consumption by radiation of a medium. The model is applied within a continuous pulse of radiation. We define the FLASH factor as the dose associated with the DNA-damage induced by classical treatments, and the sparing effect as the ratio of the dose delivered to the FLASH factor (i.e. 1 if no flash effect present and lower in case of sparing. We apply the model to the seminal in-vitro data (Town (Nature 1967), and more recent in-vivo data on cognitive performance (recognition rates) in irradiated mice (Montay-Gruel, Radiotherapy Oncology 2019). The susceptibility of the FLASH-effect to dosimetric errors is assessed by applying the model in a single pulse regimen of 15Gy delivered in 3.5μs. This is done by varying the dose by 10% at different initial oxygen levels with a range of 0 to 30 mmHg pO2 and estimating the sparing effect at each level. The cell data (Town, 1967) present the cell survival as a function of dose in multiple and single pulsed irradiation. The model follows the data well up to 20Gy, after which the model underestimated the sparing effect. A critical parameter for good agreement was the oxygenation level. The in-vivo data (Montay-Gruel 2019) applies pulsed radiation at different dose rates. The model, using pulse height only, is well correlated with the experimental data, comparing recognition rates with induced DNA-damage. The correlation ρ = 0.88 is shown to be highly significant (p<0.01) using Spearman Ranking. Finally, the error-analysis shows a dependence of the estimate of tissue sparing with respect to the initial oxygen level. A maximum difference is observed at a 23mmHg pO2 initial oxygenation in the tissue. A 10% reduction in dose results in a reduction in sparing from 0.78 to 0.86. In the region of maximal sparing the difference is smaller (0.59 to 0.6). We have shown that a model based on oxygen consumption can be used to address the behavior of FLASH radiation therapy in-vitro and in-vivo. The simplicity of the model allows extension to use in other modalities like protons and carbon-ions. In addition, allowance for different pulse shapes and pulse repetitions are easily incorporated. The next step is the inclusion of this model-based approach in a treatment planning system.