Abstract Book

S166

ESTRO 37

Material and Methods 143 head and neck cancer patients planned to receive a prescribed dose of 70 Gy in 35 fractions with daily CBCT guidance were included in this retrospective study. The actual delivered dose was estimated by 1) deforming the planning CT to the anatomy of the daily CBCT, 2) recalculating the treatment plan on this synthetic CT, 3) deforming the recalculated dose back to the planning CT, 4) accumulating all daily dose distributions. The parameters (m, TD 50 , n) of the Lyman-Kutcher-Burman (LKB) NTCP model were estimated with the maximum likelihood method by relating the equivalent uniform dose (EUD) of organs at risk (OAR) (oral cavity, contralateral parotid gland, and constrictor muscle) with toxicities (oral mucositis (≥G3), xerostomia (≥G2) and dysphagia (≥G3)) respectively. Due to the prior irradiation, baseline toxicity and missing contours, 11, 16, and 34 patients were excluded from the analysis for oral mucositis, xerostomia, and dysphagia respectively. To compare the model parameters, we performed the deviance test. To evaluate the performance, receiver- operator characteristic (ROC) analysis was performed and the area under the curve (AUC) was compared with z- test. The statistical significance was set to p<0.05. Results In this cohort the incidence for oral mucositis (≥G3), xerostomia (≥G2) and dysphagia (≥G3) was 25%, 26.8% and 30.3% respectively. The median EUD difference (0.31 Gy) between planned and delivered dose to the OAR were significantly larger for patients with dysphagia ≥Grade 3 than without (Figure 1). For oral mucositis and xerostomia, no significant differences were observed. Since there were no significant differences between 2 parameters (n=1) and 3 parameters model, the simpler model using mean dose (n=1) was used for the rest of the analysis. The delivered dose provided an equal or better fit for all toxicities and the steepness of the NTCP curve was equal or higher (smaller m) (Table 1). These differences, however were not significant. Similarly, no significant differences were found in AUC.

Conclusion Modest differences between planned and delivered dose were found in a relatively large patient cohort. Delivered dose yielded steeper NTCP models with improved fits but differences were not significant. Moreover, the difference between the models were small. Consequently, the potential clinical relevance of NTCP models based on accumulated dose for oral mucositis, xerostomia and dysphagia in head and neck cancer radiotherapy is likely to be limited. PV-0316 Development of a prediction model for unfavourable aesthetic outcome after breast- conserving therapy I. Kindts 1 , G. Defraene 2 , A. Laenen 3 , S. Petillion 1 , E. Van Limbergen 1 , T. Depuydt 1 , C. Weltens 1 1 University Hospitals Leuven, Department of Radiation Oncology, Leuven, Belgium 2 KU Leuven – University of Leuven, Department of Oncology - Experimental Radiation Oncology, Leuven, Belgium 3 KU Leuven – University of Leuven, Leuven Biostatistics and Statistical Bioinformatics Centre L-Biostat, Leuven, Belgium Purpose or Objective The aim of the present study was to develop a normal tissue complication probability (NTCP) model for late unfavourable aesthetic outcome (AO) after breast- conserving radiation therapy. Material and Methods The BCCT.core software was used to evaluate the AO of patients treated at one institution with breast-conserving radiation therapy, based on standardized photographs. Follow up periods ranged from two to six years. Literature data have shown there is no measurable change in AO measured by BCCT.core two to six years after surgery. Individual radiotherapy plans were assessed and the radiotherapy doses were recalculated to biologically equivalent doses using an α/β-value of 3.6 Gy. Uni- and multivariable logistic regression analysis was performed to study the predictive value of clinicopathological and dosimetric variables for an unfavourable AO. The Lyman Kutcher Burman (LKB) model was fit to the data with dose modifying factors (dmf). Model performance was assessed with the Area Under the Curve (AUC) of the receiver operating characteristic curve and bootstrap sampling. Calibration was evaluated using a Hosmer-Lemeshow test for the multivariable logistic regression model and a comparison with the LKB model was done with the loglikelihood. Results Forty-four of the 121 analysed patients (36%) developed unfavourable AO. In the optimal multivariable logistic regression model a larger breast volume receiving ≥55 Gy (V55), a seroma and an axillary lymph node dissection (ALND) were independently associated with an unfavourable AO. In individual cases, the probability of unfavourable AO can be calculated using the following formula: NTCP = 1 / (1+e -S ), where S = -2.6759 + (0.0574 x V55) + 1.5546 (if seroma) + 1.2029 (if ALND) (Figure). Bootstrap validation resulted in AUC 0.75 (95% CI 0.64; 0.85), the Hosmer- Lemeshow test was not significant and the loglikelihood was -51.8. The optimal LKB model parameters were EUD 3.6 (50) = 63.3 Gy (95% CI 59.2;68.1), n = 1.00 (95% CI 0.62;1), m = 0.23 (95% CI 0.16;0.37), dmf (seroma) = 0.83 (95% CI 0.73;0.95) and dmf (ALND) = 0.84 (95% CI 0.75;0.94). Bootstrap validation resulted in AUC 0.74 (95% CI 0.61; 0.83). The loglikelihood was -54.3, this is a lower value than in the multivariable logistic regression model and indicates a better calibration of the multivariable logistic regression model.