A multilevel probabilistic dose response model for radiation-induced Harderian gland tumor prevalence–single beam and mixed field effects

Abstract

Space radiation exposure is one of the main risks faced by crewmembers during exploration missions. Currently, the greatest uncertainty in radiation-induced cancer risk estimation for spaceflight missions is in the calculation of the quality factor (QF) largely due to limited ground-based experimental data for the vast range of radiation types found in space. Such data are needed to guide and validate dose-response models and subsequent calculations of relative biological effectiveness (RBE) used for QF assignment. In this work, a dose-response model is developed on the basis of probabilistic arguments and considers processes occurring at the cellular and tissue levels that ultimately result in radiation induced tumorigenesis. Probabilistic methods are extensively relied upon within the model development, so that the dose dependence is ultimately guided by the probability laws rather than any a priori assumption, such as linearity. Model components account for tumor occurrence due to targeted effects (TE), including cell sterilizations, and non-targeted effects (NTE). The model is developed for both single mono-energetic ion beam exposures and mixed field exposures (i.e., involving more than one particle type and energy). The model is calibrated with global fittings to the Harderian gland tumor incidence observations for acute irradiation from single beams reported in Alpen et al. (1993, 1994) and Chang et al. (2016). Both the TE-only model (with NTE opted out) and TE-NTE model fit the observations well, with the latter slightly better for the experiments with high linear energy transfer radiations. Importantly, the model parameters calibrated using single-ion data are then used to make tumorigenesis predictions for mixed field experimental data, as measured by Huang et al. (2020). It is found that the model can predict mixed field irradiations given sufficient single-ion data to allow robust parameter calibration. Mixed-field RBE calculations are performed with the validated model and compared to results obtained with linear dose-response and simple additivity models.