Abstract Book

S1097

ESTRO 37

small, they increased when approaching the particle range. The effect of varying the parameters on the SOBP increased with increasing variation, especially when varying r d and R n which completely altered the SOBP (Figure 1B). However, the resulting percentage change in the dose distributions was generally less than the percentage change of the parameters (Table 1). Similar results were obtained for single SOBPs at other dose levels. Also, with two opposing SOBPs the dose gradients disappeared, giving a more homogeneous dose to target. When applying parameters related to the other cell-lines, there was an approximately constant difference in the z- values from the HSG z-values, throughout the beam (Figure 1C). The SOBPs estimated applying these parameters showed in general a higher dose than the HSG SOBP, with a slight increase in dose towards the particle range (Figure 1D).

correlations between toxicity and different dose-volume parameters were analyzed. Results The incidence of late grade 1 and grade 2 GI toxicity in this patient group was 31 % and 12.5 %, while that of late GU toxicity of grade 1 and grade 2 was 32 % and 11 %, respectively. No grade 3 toxicity was recorded. For all endpoints, a statistically significant correlation between total dose to the GTV and toxicity could be established with a multivariate analysis. On the other hand, correlations between side effects and selected dose- volume parameters were not significant. In all risk groups, the mean NTCP for GI toxicity resulted to be lower than the recorded incidence while for GU toxicity the predictions agreed with the observed values within the standard deviation. Conclusion A retrospective analysis of the NTCP model predictions for the prostate cancer patients considered in this work revealed an underestimation of the late GI effects and a good agreement, although within a large spread of the values, with the observed GU toxicity. A preliminary analysis of intra-fractional internal organ movements revealed a negative effect on the NTCP prediction in the case of systematic reduction of the bladder volume or increase of the rectum volume that need further investigations. EP-2012 Sensitivity study of the Microdosimetric Kinetic Model input parameters for carbon ion radiotherapy T.J. Dahle 1 , G. Magro 2 , C.H. Stokkevåg 3 , K.S. Ytre- Hauge 1 , A. Mairani 2,4 1 University of Bergen, Dept. of Physics and Technology, Bergen, Norway 2 Centro Nazionale di Adroterapia Oncologica- CNAO, Medical Physics, Pavia, Italy 3 Haukeland University Hospital, Dept. of Oncology and Medical Physics, Bergen, Norway 4 Heidelberg Ion Beam Therapy Centre- HIT, Physics, Heidelberg, Germany Purpose or Objective An advantage of carbon ion therapy is the possibility of an increased relative biological effectiveness (RBE) in the tumor volume, compared to the RBE in the surrounding healthy tissue. The RBE is accounted for clinically by optimizing the RBE-weighted dose using biophysical models. In Japan, treatment planning for carbon ions is based on the Microdosimetric Kinetic Model (MKM) with parameters optimized for HSG cells. In this study we investigated the sensitivity of the MKM to variations in the model parameters, as the accuracy of the applied RBE model and its input parameters may influence the delivery of (homogeneous) RBE-weighted dose to the target. Material and Methods The MKM is based on estimations of specific energy (z) in small volumes called domains. To enable Monte Carlo (MC) estimations of z, a table connecting z to the kinetic energy of the particles was generated, using the Kiefer- Chatterjee track structure model with saturation corrections. The input parameters were the linear- quadratic model parameter β, the nucleus radius (R n ) and the domain radius (r d ). Based on this, several spread out Bragg peak (SOBP) scenarios were simulated in the FLUKA MC code. The MKM parameters (α, β, R n and r d ) for HSG cells were then varied one at a time by ±{5, 25, 50}%, and in addition changed to related parameter values of other cell-lines (V79, T1 and CHO). The resulting z-tables and SOBPs were compared to the nominal HSG z-table and SOBP. Results Variations in r d had the largest impact on the z-values (Figure 1A). While variations in z due to R n and β were

Conclusion Variations in input parameters altered both the shape and magnitude of the SOBP, particular for variations of R n and r d . Applying parameters for another cell-line gave a