ESTRO 38 Abstract book

S511 ESTRO 38

toxicity. Robustly optimized head and neck plans using SPRs calculated using DECT and SECT were evaluated for OAR sparing and impact of inter-fraction setup variation. Material and Methods Ten post-operative head and neck patients (2 with SCC of the tongue, 1 parotid, 1 submandibular gland, 5 tonsil, and 1 unknown primary) were used for this study. Patients received a SECT scan followed by a DECT scan in sequential mode (80 kVp and 140 kVp) whereby ρ e and Z eff images were generated by the scanner software for SPR calculation. Both SECT and DECT calculated SPR images were used for robust optimization and dose calculation. Each plan consisted of 3 fields using multi-field optimization (MFO) to a CTV prescription of 60 or 63 Gy in 30 fractions. The proton plans were robustly optimized for CTV target coverage on SECT (3.5% range uncertainty) and DECT (2% range uncertainty) scans with 3 mm isocenter shifts. Comparison of treated volume outside of CTV- 50%, 70%, and 90% isodose volumes (minus the target volumes) were performed between SECT and DECT plans. Verification scans acquired during the course of treatment were used for forward calculation and evaluated for DECT plan robustness using a 1.5 mm isocenter shift and 2% range uncertainty. The 1.5 mm isocenter shift robustness parameter for the evaluation scans reflect the uncertainty in imaging and treatment isocenter coincidence as well as user dependent variability in image registration. Results The average reduction in the 50%, 70%, and 90% isodose volumes were 6.8%, 7.3%, and 13.3% respectively. An average decrease in esophagus max of 3.0 Gy (range -8.6 to 1.4 Gy), esophagus mean 0.6 Gy (-1.8 to 0.47 Gy), constrictor mean 1.2 Gy (-2.7 to 1.1 Gy), larynx mean 1.1 Gy (-2.7 to 0.6 Gy), and parotid mean 0.3 Gy (-3.9 to 1.7 Gy) were achieved with DECT plans. CTV coverage (at least 95%-95%) was maintained on each patient’s verification scan in the second-to-worst case plan robustness analysis. Conclusion Using DECT for SPR calculation allows for reduced range uncertainty and margins leading to reduction of dose to regions outside target compared to SECT. Plans with reduced margins of 2% range uncertainty and 3 mm isocenter shifts were robust against inter-fraction setup variations.

Poster: Physics track: Radiobiological and predictive modelling, and radiomics

PO-0947 The impact of dose deviations arising within the dosimetry chain on clinical outcomes M. Bolt 1 , A. Nisbet 2 , T. Chen 3 , C. Clark 4 1 Royal Surrey County Hospital, Radiotherapy Physics, Guildford, United Kingdom ; 2 University of Surrey, Physics, Guildford, United Kingdom ; 3 University of Surrey, Chemical and Process Engineering, Guildford, United Kingdom ; 4 National Physical Laboratory, Radiation and Dosimetry, Teddington, United Kingdom Purpose or Objective Delivered radiotherapy dose may be compared between clinics due to the traceability of dose between Primary Standards Laboratories (PSL). Uncertainties arise at each point in the calibration chain and thus the actual delivered dose can deviate from that desired. These uncertainties have been quantified for each step; calibration transfer to a secondary standard instrument, transfer to field instruments, and subsequent QA measurement tolerances. Radiobiological modelling has been used to predict the clinical impact of these uncertainties. Material and Methods Uncertainty in the initial transfer of the calibration from PSL to the clinic was determined from on-site audits performed by the PSL over 2 decades and is normally distributed with a SD of 0.7% [1]. Data from over 24,000 multi-centre beam output measurements was used as a basis of the uncertainty following calibration and had a 0.7% SD [2]. Combining these uncertainties with daily fluctuations (0.2% SD) gives the overall measured uncertainty within the calibration chain of 1.0% SD. Linear-Quadratic (LQ) and Lyman-Kutcher-Burman (LKB) models were implemented to model the effect of systematic and random deviations in delivered dose. Modelling was developed for prostate (10yr bPFS and grade 2 rectal bleeding) and head and neck (2yr survival and xerostomia induction) cases. The LQ model was used to calculate individual patient response and aggregated to provide population estimates. The clinical cases assessed cover a wide range of dose-response (even some prostate cancers have steep dose response [3]). Results A systematic dose shift of +2% through the course of treatment was estimated to change TCP by between 6.1- 7.0% for the populations. Table 1 shows results for a range