ESTRO 37 Abstract book

S756

ESTRO 37

1mm (range-6 to 4). The standard deviation of the σ and ∑ setup errors are displayed in Table 2. Of all 41 patients, just 3 (7%) pts were categorised as high risk intrafraction movement group and required improvement actions during most days of treatment.

acute esophagus toxicity (AET) for concurrent chemoradiation (CCRT) for NSCLC patients were used to validate the ETR of AET before/after 1 st June 2015. At this date, dose on the irradiated mediastinal lymph nodes was de-escalated from 66 Gy to 58,08 Gy (60Gy BED) in 24 fractions. Material and Methods To validate the NTCP-models, clinical data of 219 (N=101 before 1 st June 2015) inoperable NSCLC patients receiving CCRT using IMRT was used. The planned V50Gy and V60Gy and the prospectively scored grade ≥2 AET were retrieved. The grade ≥2 AET probability for the V50Gy and V60Gy was calculated as; [1/1+ exp [ -0.515 + (0.027*V50 ]] and [1/1+ exp [ -0.701 + (0.029*V60 ]]. Validity of the model was assessed with the ability to predict the number of grade ≥2 AET events (calibration) and the ability to distinguish between those who develop grade ≥2 AET from those who do not (discrimination, area under the curve (AUC)). Furthermore, sensitivity and specificity for different cut- off points were determined. Results The incidence of grade ≥2 and grade ≥3 AET, decreased between the 2 periods from 51% to 37,6% and 8.8% to 3.4% respectively. Calibration showed that the V50 overestimated the risk of grade ≥2 AET in low-risk patients while the V60 underestimated the risk of grade ≥2 AET in high-risk patients. Discrimination of both algorithms demonstrated a similar moderate fit (AUC 0.70 and AUC 0.68 for the V50 and V60, respectively). Cut-off points at 40% and 75% probability of grade ≥2 AET resulted in most favorable sensibility of 95.8% (specificity 30.1%) and 99.2% (specificity 2.1%) for V50 and V60, respectively. For the V50-model an almost similar model- fit was found (AUC before 1 st June 2015: 0.69, and AUC after: 0.70. For the V60-model the model fit decreased after dose de-escalation; AUC= 0.726 compared to an AUC = 0.616, respectively (Figure1). Conclusion RWE is a useful method to audit NTCP-models in clinical practice. De-escalation of the mediastinal dose decreased the AET. The NTCP-models to predict AET in NSCLC patients both showed good predictive accuracy. For clinical practice, the V50Gy seems to be the most stable to the dose de-escalation and sensitive without compromising safety and efficacy. EP-1384 The role of SABR for eradicative treatment of thymoma recurrences C. Menichelli 1 , C. Casamassima 2 , G. Pastore 3 , A. Fanelli 4 , V. Mazzotti 5 , E. Lombardo 6 , M. Cantarella 4 , A. Chella 7 , J. Petrini 8 1 Centro di Radioterapia - Ecomedica, Resarch Institute- Ecomedica, Empoli, Italy 2 Resarch Institute Ecomedica, Radiation Oncology, empoli, Italy 3 Research Institute, Radiation Physics, Empoli, Italy 4 Research Institute, Radiation oncology, empoli, Italy 5 Reserach Institute, Radiation Oncology, Empoli, Italy 6 Research Institute, radiaton Oncology, Empoli, Italy 7 University of Pisa, Thoracic Surgery, pisa, Italy 8 Univversity of Pisa, Medical Oncology, pisa, Italy Purpose or Objective Surgery usually represents the standard treatment of Thymoma with neoadjuvant or adjuvant chemoradiotherapy in locally advanced stages. Frequently the recurrences of disease appears on the pleura. The relative radiosensitivity of thymomas and the small tendency to systemic spread offer curative possibility to local eradicative therapies. We have evaluated in terms to local control, OS and toxicity the SABR directed to all sites of relapses. Material and Methods Between April 2014 and February 2017, 17 pts with thymoma underwent SABR for pleural relapses Median

Conclusion We can conclude that intrafraction movements are not negligible, and probably depend on many factors that are necessary to control in a second analysis. The differences between the intrafraction displacements, observed through treatment using CBCTs, seem to be clinically acceptable and are contemplated in our PTV margin. The implementation of a CBCT after fraction increases the total treatment time, but we do consider it necessary for a qualified treatment. EP-1383 The use of Real World Evidence to audit NTCP-models for acute esophagus toxicity in NSCLC patients M.H. Kwint 1 , I. Walraven 1 , M.S. Marshall 1 , M. Verheij 1 , Purpose or Objective Clinical audits of NTCP-models are ideally supported by real world evidence (RWE); real world data derived from observational studies and/or clinical registries. RWE is often criticized due to its lack of validity caused by missing data compared to results of clinical trials. By introducing electronic toxicity registration (ETR) within our institute, a more systematic and accurate recording of toxicity was implemented in clinical practice. Using ETR, toxicity predictions and dose constraints can be audited. Hence, the aim of this work is to assess the validity of the RWE toxicity registration and to show the feasibility of ETR to audit toxicity prediction models and dose constraints. As a showcase, the NTCP-models of J.J. Sonke 1 , J.S.A. Belderbos 1 , T.M. Janssen 1 1 Netherlands Cancer Institute, Radiotherapy, Amsterdam, The Netherlands