Studies of the linear energy transfer and relative biological effectiveness in proton therapy of pediatric brain tumors

Abstract

Proton therapy offers a reduction in dose to normal tissue compared to conventional photon-based radiotherapy. This is of particular benefit for pediatric patients as the majority are expected to become long-term survivors. Children are therefore often referred to proton therapy in order to reduce the risk of radiation induced side effects. Protons are also slightly more biologically effective compared to photons, quantified by the clinically applied relative biological effectiveness (RBE) of 1.1. However, both experimental and clinical data points to a variable RBE, which depends on tissue type, dose level, biological endpoint, and the linear energy transfer (LET). Multiple variable RBE models have therefore been developed with the aim of quantifying the RBE variation.

Brain tumor patients are often at high risk of radiation damage to the brainstem - a vital organ where injury can lead to devastating side effects. Minimizing doses to the brainstem has therefore a high priority during treatment planning. However, the brainstem may also be adversely affected by the LET and variable RBE, factors that are not explicitly accounted for in routine proton therapy. In this PhD project, for both double scattering and intensity modulated proton therapy (IMPT), the LET and variable RBE in the brainstem for pediatric brain tumor patients were studied using the FLUKA Monte Carlo (MC) code.

In the first part of this project, the LET and RBE in the brainstem were studied for different tumor locations relative to the brainstem. Furthermore, techniques for reducing the LET in critical organs by changing the treatment field setup were explored (Paper I). Mean LET values in the brainstem more than doubled depending on the tumor location (3.2-6.6 keV/μm), however, the location with the highest brainstem LET values also had the lowest variable RBE-weighted mean dose in the brainstem (1.8-54.0 Gy(RBE)). Changing treatment field angles reduced the mean LET in the brainstem by 32%, however, with slightly increased brainstem dose. The results demonstrate that the LET and variable RBE-weighted dose are strongly influenced by tumor location and field configuration, and that both LET and variable RBE-weighted dose must be carefully considered when altering treatment plans.

In the second part, multiple variable RBE models in treatment for pediatric brain tumors were investigated. The spatial agreement of isodose volumes from the models relative to the RBE of 1.1 were compared, focusing on the full brainstem and brainstem substructures (Paper II). Application of different model specific parameters were also explored. The RBE-weighted dose calculated from RBE models was highly dependent on the applied parameters, and also differed across models. Furthermore, the spatial agreement between different models decreased rapidly for higher doses, illustrating that the RBE effect is most critical at high doses and low volumes, where dose constraints commonly are applied. Hence, using RBE models in clinical settings requires model specific dose constraints.

The majority of follow-up data from proton therapy come from patients treated with double scattering (DS) proton therapy. Therefore, a DS nozzle was implemented in the FLUKA MC code in order to obtain LET and variable RBE for previously treated patients (Paper III). After calibration, excellent agreement between measurements and MC simulations was achieved with range differences of spread-out Bragg peaks generally below 1 mm and lateral penumbra differences less than 1 mm. Recalculation of dose distributions in FLUKA were compared to original patient doses from the treatment planning system, with dose differences below 2%. LET and variable RBE were furthermore obtained for these patients.

In the final part of this project, the DS nozzle implementation was used to recalculate 36 pediatric brain tumor patients in a retrospective case-control study where nine patients had experienced symptomatic brainstem toxicity. Differences in LET and variable RBE-weighted dose between cases and controls were examined for the full brainstem as well as multiple brainstem substructures. Median and maximum LET were on average higher for cases vs. controls for all substructures, with the highest difference in median LET of 15% in one of the substructures. Average differences between cases and controls increased for variable RBE-weighted doses compared to a fixed RBE of 1.1. While there was large interpatient variability for both LET and variable RBE-weighted doses, the average higher LET to the brainstem could be a contributor to brainstem toxicity. The results warrant individual assessment of LET/RBE for patients at risk of brainstem toxicity.

Overall, this thesis has shown that elevated LET and increased RBE may occur in the brainstem for pediatric patients with brain tumors which could further contribute to brainstem toxicity. Clinical implementation of LET and variable RBE-weighted dose calculation is therefore well justified.