dc.description.abstract
Purpose: At the MedAustron (MA) Ion Therapy Centre, medical commissioning of Carbon ion beams was performed at the fixed Horizontal Beam Line in Irradiation Room 2 (IR2HBL). Carbon ion beams in the energy range of 120402.8 MeV/u (corresponding to 29.2270 mm range in water) and a spot size of at least 6 mm FWHM in air are available. In addition to the advantageous depth-dose profiles (compared to photons) and lateral scattering (compared to protons), Carbon ions offer a larger efficacy for tumor cells killing due to the enhanced biological effect or Relative Biological Effectiveness (RBE). Non-elastic nuclear interactions of the primary Carbon ion projectiles with the target nuclei result in a reduction of primary ions and a build-up of lower-Z fragments with longer ranges, causing the characteristic fragmentation tail beyond the Bragg peak. At MA, the treatment planning system (TPS) RayStation (RS) v8B (RaySearch Laboratories RSL, Sweden) has been commissioned. For RBE-weighted dose computation, the Local Effect Model version 1 (LEM I) is used. In order to compute the biological effect, the LEM I requires the energy spectrum of all particles and the physical absorbed dose as input parameters. RSL have used the FLUKA Monte Carlo (MC) code to compute particle energy spectra kernels of mono-energetic Carbon ion beams as input in the pencil beam dose engine (PBv3.0). Interactions with nozzle elements are handled by offsetting and weighting the spectra kernels, in order to reproduce the energy distribution at the nozzle exit. However, as the MA nozzle is not available in FLUKA, the interactions of the Carbon ion beams with the nozzle elements cannot be considered explicitly. This work aims to study the impact of beamline specific particle energy spectra on the RBE-weighted dose for clinical treatment plans. Materials and methods: At MA a TPS independent MC particle transport code (Gate v8.2 Geant4 v10.3 patch 03) was used to simulate the entire beam line (IR2HBL) with the detailed description of all nozzle components. First, based on data acquired during the commissioning phase, the optics parameters, the energy, and the energy spread were tuned in Gate to reproduce the beam characteristics of IR2HBL. The hadron physics builder was chosen, comparing simulated integrated radial profiles as function of depth (IRPD) with profiles acquired with three differentsized plane-parallel ionization chambers PPIC (diameter of 39.6 mm, 81.6 mm, 147 mm) in water. Following a similar approach as RSL in FLUKA, particle energy spectra were simulated with GATE for single energy beams (ranging from 120 to 402.8 MeV/u in 5 MeV/u steps). Based on the two sets of spectra, two clinical beam models were created in RS: one beam model containing the RSL-FLUKA pre-generated particle spectra and one containing the MA-Gate beamline-specific particle spectra. The impact of the different particle energy spectra (RSL-FLUKA vs. MA-Gate spectra) on the RBE-weighted biological dose was evaluated for differently sized and shaped targets in water and for some clinical plans with and without range shifter. Results: Among the different hadron physics configurations, “Shielding” (containing the Quantum Molecular Dynamics model) agreed within 5% with the measured integrated depth dose profiles in the plateau and the Bragg peak region. Larger local deviations (up to 20%) were found for all the models in the fragmentation tail, but in these regions, less than 4% of the energy is deposited in comparison to the Bragg peak, therefore they are clinically not relevant. As a result, the reference hadron physics builder “Shielding” was selected for the simulation of the particle energy spectra. RSL-FLUKA pre-generated particle spectra were benchmarked against the MA-Gate beamline-specific spectra. The MA-Gate and the RSL-FLUKA fluence agreed well for the main contribution, the primary Carbon ions. However, the fluence at energies below 1 MeV/u was considerably lower in GATE for all secondary particles and mainly for the proton and helium fluence components. The dose contribution of low energy particles in the spread-out Bragg peak (SOBP) is low, but the impact on RBE-weighted dose was quantified as the RBE is non-linearly increasing with decreasing energy. Single energy layer results revealed that the compared RBE-weighted dose generated by the two different beam models, the MA-GATE and RSL-FLUKA models differ about less than 3% in the entrance region till the Bragg peak, whereas during the Bragg peak fall off region the deviation increases up to 5% and in the fragmentation tail differences of up to 15% were found. For the boxes in water three regions were investigated. The deviations found in the plateau region were less than 0.5%, in the target region less than 1% and in the fragmentation tail less than 5%. The results from the fragmentation tails were tolerable because the fragmentation tail is a low dose region, with a small absolute but a high relative dose deviation. In the five clinical cases, we investigated the dose distributions which showed differences of up to 3% restricted to local spots. The target volume dose deviations were found to be up to 2%, the maximum dose difference in the organs at risk in the clinical plans was 5.9%, although this value and other higher deviations were found when the expected (prescribed) dose for organs at risk was found to be far above the observed dose. Conclusion: The correct prediction of particle energy spectra at a certain depth in tissue for Carbon ion beams is essential to assess, within acceptable clinical tolerances, the RBE-weighted dose in the patient. Therefore, it is crucial to evaluate the impact of different MC codes, non-elastic models, and nozzle components on the RBE-weighted dose for Carbon ions. In this study, an independent MC code was used to simulate the full nozzle for the IR2HBL with Carbon ions. A selection of the most suitable hadron physics configuration was made based on depth dose profiles acquired with PPIC at different radii. In total no relevant clinical differences in the RBE-weighted dose comparison in the TPS RS were found neither in the target geometries in water nor in the clinical cases in the delivered dose produced by the beam models RSL-FLUKA and MA-GATE. Consequently, Geant4/GATE may be used to independently validate Carbon ion beams for commissioning of a Carbon ion beam model and the generation of the particle energy spectra required for the LEM I model and further the RBE-weighted dose computation.
de
