ESTRO 38 Abstract book

S542 ESTRO 38

and shape of target volume. In whole breast RT with axillary lymph node involvement the clinical rotational setup errors can result in over 6 mm translational shifts at the edge of the PTV. PO-0988 CBCT-based library of plans approach in gastric cancer radiotherapy: proof of concept M. Bleeker 1 , K. Goudschaal 1 , A. Bel 1 , J. Sonke 1,2 , M.C.C.M. Hulshof 1 , A. Van der Horst 1 1 Amsterdam UMC- University of Amsterdam, Radiation Oncology, Amsterdam, The Netherlands ; 2 Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective Current protocols for gastric radiotherapy (RT) use one treatment plan only. A library of plans for gastric cancer may ensure adequate irradiation by adapting the treatment to large day-to-day anatomical changes in size, shape and position. Here, we assess the feasibility of CBCT-guided gastric cancer adaptive RT (ART) with a library of plans based on the planning CT (pCT) and the daily CBCTs of the first week of RT. Material and Methods Three patients, who had received single-plan pre- operative RT within the CRITICS-II trial (NCT02931890; 45 Gy; 25 fractions; PTV margin = 10 mm) with daily CBCT- guidance, were included. For this retrospective study, all CBCTs were rigidly registered to the pCT based on bony anatomy and if needed corrected in longitudinal direction for diaphragm position (Velocity 4.0, Varian). One expert radiation oncologist manually delineated CTV (i.e. entire stomach and regional lymph nodes) according to trial protocol on pCT and the first 5 CBCTs (Fig 1A). For delineation on CBCT, the CTV_pCT was copied and subsequently adjusted. From these 6 CTVs, we created 3 Plan Selection Volumes (PSVs) (Fig 1B): PSV_large = union of all 6 CTVs + 3 mm, PSV_medium = volume occupied by 4 of 6 CTVs (i.e. volume with at least 4 overlapping CTVs) + 3 mm, and PSV_small = smallest delineated CTV + 3 mm; 3 mm was used to preserve a 7 mm PTV margin to account for remaining uncertainties. A specially trained RTT selected the most appropriate PSV for the remaining 20 fractions (i.e. fractions 6-25). The selected PSV must encompass the CTV as visible on the CBCT.

Fig 1. The directions of the rotations investigated in this study. Results The mean absolute translational shifts from skin marker- based patient setup were 4 ± 3 mm, 3 ± 3 mm and 4 ± 4 mm in anterior-posterior, left-right and cranio-caudal directions, respectively. The residual rotational errors were on average 0.0 ± 1.4° for COR, 0.2 ± 1.6° for SAG and 0.2 ± 1.9° for TRA directions (Fig 2). The mean absolute rotations were 1.0 ± 0.9° for COR, 1.2 ± 1.0° for SAG and 1.4 ± 1.3° for TRA directions, respectively. Overall, 35% of the residual rotational errors were larger than 2°. In 244 fractions (14%) the rotational error was over 3°. A three-degree rotational error would result in a translational shift of 4 mm at the edge of a typical breast cancer PTV (radius 8 cm), and for a PTV with axillary lymph nodes involved (radius up to 12 cm), the translational error would increase up to 6 mm. Modest but statistically significant correlations where found between patient age ( r =0.27, p =0.01, smaller errors with younger age) and the use of DIBH ( r =0.23, p =0.02, smaller errors with DIBH) and the rotational errors.

Fig 2 . Histogram of the residual rotational errors in coronal, sagittal and transversal directions for all investigated setup images. Conclusion With online patient position correction methods, the translational setup deviations can be reduced to negligible values. However, the rotational residual setup errors should also be accounted when determine the PTV margins. Furthermore, the PTV margins taking into account the rotational errors should depend on the size