Abstract Book

S1092

ESTRO 37

Purpose or Objective To analyze and compare normal tissue complication probability (NTCP) models in their ability to predict late urinary toxicity at 5 years follow-up, in a series of patients having received 3D conformal radiotherapy (3DCRT) or intensity-modulated radiotherapy (IMRT) for prostate cancer. Material and Methods Clinical data and dose-volume histograms (DVH) were collected in a series of 434 patients, following the same dose-volume recommendations (from the GETUG group). Bladder toxicity endpoints were: ≥Grade 2 (LENT/SOMA grading system) and bleeding (at least one episode). The correlations between toxicity and patient characteristics or DVH were investigated. Parameters of 6 NTCP models (LKB, Logit-EUD, Poisson-EUD, Källman, Schultheiss and Parallel) were identified at 5 years follow-up. Toxicity prediction performance of each model was compared by Receiver Operating Characteristic methods. Results LKB, Logit-EUD, Poisson-EUD, Källman, Shcultheiss and Parallel models predicted late bladder toxicity (≥Grade 2) and bladder bleeding with similar capabilities (area under curve ranges from 0.661 to 0.665 and 0.623 to 0.627, respectively). Multivariate analysis shown that: no DVH bin was significant when considering any of the six NTCP models at the same time; NTCP values and EUD were also stronger predictors than the maximum dose and the total dose.

Conclusion Among the methods studied, there is a clear trade-off between accuracy and time consumption. For MHD estimation, the %Heart method offers a good compromise and is the quickest way of estimating the MHD within 1.6 Gy when 2D simulation films are available. The RepCT and the Navigator methods are also accurate but more time consuming, therefore more appropriate for smaller patient cohorts. The Phantom method is arguably the most practical option for large patient cohorts (e.g. Childhood Cancer Survivor Study). EP-2005 NTCP model to predict late urinary toxicity after prostate cancer radiotherapy J. Zhu 1 , A. Simon 2 , J. Ospina 2 , A. Bossi 3 , C. Chira 4 , K. Gnep 4 , V. Beckendorf 5 , R. De Crevoisier 2 1 Shandong Cancer Hospital affiliated to Shandong University, Department of Radiation Oncology Physic and Technology, Jinan, China 2 INSERM U1099, Laboratoire Traitement du Signal et de l'Image, Rennes, France 3 Institut Gustave Roussy, Département de Radiothérapie, Villejuif, France 4 Centre Eugène Marquis, Département de Radiothérapie, Rennes, France 5 Centre Alexis Vautrin, Département de Radiothérapie, Nancy, France

Conclusion All the six NTCP models present a similar significant capabilities at predicting late bladder toxicity (≥Grade 2) and bladder bleeding with the identified parameters. Both NTCP and DVH parameters are useful to assess late bladder toxicities. EP-2006 Normal-tissue toxicity following gastric cancer radiotherapy with photon- or scanned proton- beams G.A. Mondlane 1 , A. Ureba 1 , M. Gubanski 2 , P. Lind 2,3 , A. Siegbahn 1 1 Stockholm University, Department of Physics, Stockholm, Sweden 2 Karolinska University Hospital, Department of Oncology and Pathology, Stockholm, Sweden 3 Södersjukhuset, Department of Oncology, Stockholm, Sweden Purpose or Objective The aim of this study was to evaluate the potential of scanned proton-beam therapy to reduce the normal- tissue toxicity after gastric cancer (GC) treatment, compared to photon-beam therapy.