ESTRO 36 Abstract Book

S106 ESTRO 36 _______________________________________________________________________________________________

optimization have been described in literature. ‘Minimax’ robust optimization is a relatively straightforward implementation and is currently incorporated in several treatment planning systems that are commercially available. During minimax robust optimization, dose-influence matrices are typically calculated for the nominal scenario (without treatment errors) and for a number of user-defined error scenarios, and are subsequently used to optimize worst-case values. The user can generally specify the number of included error scenarios and the magnitude of the treatment errors accounted for. In this way, one can account for errors in patient setup and in particle range, and, in some implementations, for anatomical changes. The characteristics and practicalities of minimax robust optimization in intensity-modulated proton therapy (IMPT) for oropharyngeal cancer patients will be addressed in this presentation: 1. Robustness recipes: Which robust optimization settings (i.e. error scenarios) should be used for given population- based distributions of setup and range errors (systematic/random), in order to obtain adequate clinical target volume (CTV) coverage in oropharyngeal cancer patients? Available robustness recipes differ between patients with unilateral or bilateral tumors and suggest that setup errors and range errors can be accounted for independently. 2. The price of robustness: What does robustness cost in terms of dose to organs-at-risk (OARs)? An investigation on the impact of the degree of robustness (i.e. magnitude of the included error scenarios) on OAR doses and resulting normal-tissue complication probabilities showed that setup robustness had a substantially larger impact than range robustness. This suggests that minimizing setup errors should be given a higher priority than minimizing range errors, in IMPT treatments for oropharyngeal cancer patients. 3. Minimax robust optimization to account for anatomical uncertainties. Anatomical robust optimization can effectively deal with changes in nasal cavity filling, providing substantially improved CTV and OAR doses compared with the conventional margin-based approach. Future investigations should reveal whether minimax robust optimization can also be used to account for other anatomical changes in oropharyngeal cancer patients. J.C. Lindegaard 1 , A. Ramlov 1 , M. Assenholt 1 , M. Jensen 1 , C. Grønborg 1 , R. Nout 2 , L. Fokdal 1 , K. Tanderup 1 , M. Alber 3 1 Aarhus University Hospital, Department of Oncology, Aarhus C, Denmark 2 Leiden University Medical Center, Department of Radiation Oncology, Leiden, The Netherlands 3 Heidelberg University Hospital and Heidelberg Institute for Radiation Oncology HIRO, Department of Radiation Oncology, Heidelberg, Germany Definitive radiotherapy in locally advanced cervical cancer (LACC) often includes boosting of multiple pathological pelvic nodes. The simultaneous integrated boost (SIB) technique delivered by intensity modulated radiotherapy (IMRT) or volumetric arc therapy (VMAT) is increasingly being used as recent studies have shown excellent nodal control with a boost of 55-60 Gy. However, nodal boosting on top of elective whole pelvic radiotherapy at 45-50 Gy invariably causes collateral higher dose to especially bowel and pelvic bones, as metastatic regional nodes in LACC are most often situated in the retroperitoneal lymphatic space close to both bowel loops and the pelvic wall. This dilemma may be even worse in situations where para-aortic nodes are encountered and require irradiation. At present no consensus exists on the required margin for nodal boosting by SIB, but margins of 5-10 mm from the SP-0211 Clinical implementation of coverage probability planning in cervix cancer

gross tumor volume of the node (GTV-N) to the nodal planning target volume (PTV-N) have been reported. Since the diameter of pathological nodes (GTV- N) most often is about 10-20 mm, SIB dose planning using a classical PTV concept of a dose plateau with full PTV-N coverage will entail a relatively large volume being treated to high doses compared to the actual GTV-N volume. In addition, the robustness of SIB being embedded in the 45-50 Gy irradiation of the whole pelvis is not fully utilized. Coverage probability treatment planning (CovP) has previously been shown to provide robust dose escalation for IMRT of prostate cancer with overlapping PTV and rectum planning volume as well as superior patient specific small bowel planning volume allowing for tighter OAR margins with for instance para-aortic radiotherapy. Reduction of the dose at the perimeter of the PTV-N could therefore be considered by employing coverage probability dose planning (CovP) for SIB in LACC. With CovP local weights for each voxel are being used to create a dose gradient at the edges of PTV-N according to the presumed probability of finding the nodal target at this coordinate in the treatment room. The shape of the fall- off is based on assumptions about the position error of the GTV-N. CovP has recently been implemented in the prospective international multicentre EMBRACE II study for SIB planning of nodal boosting in LACC (www.embracestudy.dk). Clinical validation and implementation of CovP treatment planning in LACC was performed at Aarhus University Hospital in 2015 as a preparation for the Embrace II study. Until then CovP had only been explored by use of experimental treatment planning systems. A first step was therefore to obtain a set of dose constraints based on a number of CovP dummy runs performed in the research dose planning software Hyperion. Based on assumptions regarding the position of GTV-N over time, the dose optimizer created a dose gradient around the CTV-N which was allowed to lie partially inside PTV-N. From these experimental CovP plans, dose constraints for use with the clinical treatment planning system Eclipse were chosen that captured the dose peak and dose gradient of the CovP dose distribution for this particular setting of SIB boosting in LACC: PTV-N D98 >90%, CTV-N D98 > 100% and a soft constraint of CTV-N D50 > 101.5% of the prescribed dose. The next step was then to analyze a number of previously treated patients with LACC. In total 25 patients with 47 boosted nodes treated with SIB delivered by IMRT or VMAT from 2012-2015 were investigated (Figure 1). Dose of EBRT was 45 Gy/25 fx with a SIB of 55-57.5 Gy depending on the expected dose from brachytherapy (BT). The planning aim was to reach D98 > 57 GyEQD2. Nodes were contoured on cone beam CT (CBCT) and the accumulated dose in GTV-N CBCT and volume of body, pelvic bones and bowel receiving >50 Gy (V50) were determined. Nearly all nodes (89%) were visible on CBCT and showed considerable regression . Total EBRT and BT D98 was >57 Gy EQD2 in 98% of the visible nodes. Compared to conventional planning, CovP significantly reduced V50 of body, bones and bowel. With CovP a new tool is available for nodal SIB in LACC allowing for controlled underdosing at the edge of the PTV. As this study is mainly based on pelvic nodes along the major vessels it is still unclear how margin reduction and CovP will perform for SIB of para-aortic nodes or nodes in the groins. Nodes in the vicinity of organs which may be displaced e.g. by the bladder or rectum may also need monitoring in terms of delivered dose and eventually plan adaptation during EBRT. However, CovP could be of interest for nodal SIB in anal, rectal, vulvar, penile, vaginal, prostate and bladder cancer. In EMBRACE II the patients are treated with a reduced PTV-N margin (5 mm), daily IGRT, IMRT/VMAT and CovP planning for SIB with planning aims presented above. With an estimated accrual of 800 patients, of which 50% will node positive disease, a