Session Item

In proton therapy, the biological dose (D_bio) is calculated by means of the relative biological effectiveness (RBE). RBE has been proposed to vary with the linear energy transfer (LET), physical dose and other factors including the oxygen level in the tumor. Models to calculate D_bio considering the effect of hypoxic cells typically estimate the oxygen enhancement ratio (OER) using the partial oxygen pressure (pO₂) and a parameter for the radiation quality such as the LET or the specific energy. Our aim was to implement and compare different OER models as input to RBE calculation and to use a Monte Carlo (MC) simulation tool to explore the impact of hypoxia on D_bio. Hypoxia PET imaging was used to estimate the pO₂ in patients in order to investigate the effects of hypoxia adaptation on proton plans.

We implemented a calculation for hypoxia-adapted D_bio, i.e., the RBE-OER-weighted dose (ROWD), based on four different approaches to estimate the OER (Wenzl and Wilkens 2011, Dahle et al. 2020, Tinganelli et al. 2015, Mein et al. 2021). The OER-calculations were combined with either a phenomenological RBE model for protons, or the microdosimetric kinetic model (MKM). First, the FLUKA MC tool was used to simulate a proton beam irradiating a virtual water phantom with different pO₂ levels. Secondly, a proton plan was made in Eclipse TPS (Varian) for a head and neck cancer case calculating the pO₂ from [¹⁸F]-EF5 PET as input to the OER estimation. For both the phantom and the patient simulations, the physical dose (D), the hypoxia-adapted linear-quadratic parameters (α_h, β_h), the dose-averaged LET (LET_d), and the OER were estimated.

Water phantom simulations using MC showed good agreement with theoretical OER models. OER estimates corresponded to clinically relevant pO₂ values and were similar between models except for the model from Tinganelli et al. 2015, which estimates approximately a 10% higher OER (Figure 1). At the same time, the models from Wenzl and Willkens 2011 and Dahle et al. 2020 are more susceptible to LET variations than the models from Mein et al. 2021 and Tinganelli et al. 2015 that particularly shows no change for the LET variations on proton LET scales.  The corresponding depth-dose curves resulted in a ROWD for all implemented models that followed the same trend with a reduction of the ROWD in the hypoxic regions (Figure 2). Preliminary results from the patient case indicates an OER range of 1 to 1.6 in the planning target volume.

We implemented a calculation of D_bio in FLUKA MC using different OER estimations and variable RBE calculations in order to perform hypoxia adaptation of proton therapy plans. Application and comparison of the models showed overall good agreement between different models, although some variation in LET sensitivity was observed. With  reliable hypoxia information, applications of OER adapted/adapting RBE models could become a useful tool for treatment plan evaluation and optimization.