ESTRO 38 Abstract book

S322 ESTRO 38

is; >73 Gy in 12%, 68 Gy in 69%, 66 Gy in 8% and 50 Gy in 9%. The dose/fraction is 2.2 Gy for the highest dose group and 2.0 Gy for the other groups. The number of fractions/week is 6 in 51% and 5 in 45% of the patients. The dose per fraction to elective volumes is 2.0 Gy or 1.52 Gy for two sequential dose plans and simultaneous integrated boost plans, respectively. A logistic regression model of grade ≥2 late xerostomia was fitted for patients who had at least one follow-up scoring late toxicity, and no missing values for the included variables. Re-irradiated patients were excluded from the analysis. Forward selection of model parameters was used and discrimination and calibration were assessed with AUC and calibration curves. An internal validation with boostrap was performed to correct the performance measure for optimism. Results Out of the 673 patients included in the study 43% had a maximum score of grade ≥2 late xerostomia. The cohort characteristics are listed in the table. The following variables were selected when fitting the model: mean dose to the total parotid volume (ipsilateral + contralateral), tumour type, gender and concomitant drug therapy. A greater odds ratio was associated with higher parotid dose, orophanynx tumours, female gender and concomitant cisplatin treatment, respectively. The optimism corrected AUC was 0.638 (95%CI: 0.597; 0.678) and the calibration slope was 0.798 (95%CI: 0.535; 1.061). The model performance did not improve when forcing the inclusion of the mean dose to each parotid separately, and the odds ratios for the two parotids were equivalent.

Conclusion The retrospective QA analysis of this prospective SBRT trial showed up to a 32% rate of dosimetric variations with potential impact on tumor control and/or toxicity profile. Use of an IMRT technique and inclusion of SV in the PTV were the major causes of protocol deviations. Integration of prospective RTQA protocols is encouraged for future PCa SBRT clinical trials to prevent and correct protocol violations before start of treatment.

Proffered Papers: PH 13: Proffered paper: Modelling toxicity

OC-0610 Modelling of xerostomia after radiotherapy for head and neck cancer: a registry study E. Onjukka 1 , C. Mercke 2 , A. Discacciati 3 , G. Alexandersson von Döbeln 2 , E. Björgvinsson 2 , H. Carstens 2 , S. Friesland 2 , G. Gagliardi 1 , C. Lenneby Helleday 2 , H. Sjödin 2 , G. Wickart Johansson 2 , I. Lax 1 1 Karolinska University Hospital, Medical Radiation Physics and Nuclear Medicine, Stockholm, Sweden ; 2 Karolinska University Hospital, Oncology, Stockholm, Sweden ; 3 Karolinska Institute, Institute of Environmental Medicine, Stockholm, Sweden Purpose or Objective The purpose of this study is to model the incidence of late xerostomia in patients treated with radiotherapy (RT) for head-and-neck (H&N) cancer, based on data from a local quality registry. Material and Methods Since 2013 patients followed up after RT for Head and Neck cancer at Karolinska University Hospital are entered into a quality registry and scored for toxicity of the skin, mucosa, larynx, and mandible, as well as xerostomia, dysphagia and trismus, according to a modified RTOG/LENT-SOMA scale. Also patient- and treatment- related factors are recorded, including RT dose/volume parameters for external-beam RT, concomitant drug therapy, performance status, gender, age, tumour location, tumour stage, HPV association and smoking status. The registry has a compliance of 70% and currently includes records of 1400 patients. The treatment technique is predominantly IMRT/VMAT (87%) with 6 MV. Prescribed dose to primary target volume

Conclusion