ESTRO 2023 - Abstract Book

S1941

Digital Posters

ESTRO 2023

1 Erasmus MC Cancer Institute, University Medical Center Rotterdam, Department of Radiotherapy, Rotterdam, The Netherlands; 2 Alma mater studiorum Bologna Institute, Physics, Bologna, Italy; 3 Alma mater studiorum Bologna Institute, Physics, Bolognait, Italy Purpose or Objective In brachytherapy (BT) planning for locally advanced cervical cancer (LACC), imposed requirements for loading patterns (e.g. limited total dwell time in interstitial needles), may impact dosimetrical quality of the plans. Previously, we developed a system for fully-automated multi-criteria optimization (MCO) of BT plans for LACC according EMBRACE II. The automated workflow generated high-quality, clinically acceptable plans controlling both dosimetry and loading pattern. In this study, automated planning was used to systematically investigate the impact of loading pattern requirements on dosimetric plan quality. Materials and Methods The first HDR BT fractions of twenty recently treated patients were used to automatically generate in total 360 plans. Investigated loading pattern parameters were 1) dwell position spacing, 2) dwell time smoothing for avoidance of large dwell time differences between neighbouring dwell positions, and 3) relative usage of intrauterine, ovoids and needles, and usage of dwell positions inside CTVHR instead of outside (‘geometrical steering parameters’). Generated plans were compared for total cumulative EQD2Gy doses (EBRT+3BT), assuming that the same BT plan was delivered in all three BT fractions. Results Fig. 1a shows population-mean dosimetrical plan parameters as a function of the spacing between neighbouring dwell positions. For the targets, bladder, rectum, sigmoid and bowel (‘MAIN’ structures), reducing the spacing from the nominal 5 mm to as low as 1 mm did only result in dosimetric changes of less than ±1.3 Gy. Enlarging the spacing from 5 to 10 mm deteriorated bladder D2cc by 3.3 Gy. The data points in the rectangle in Fig. 1b show that switching off all geometrical steering parameters (so letting this completely free in plan optimization) did for the MAIN structures at maximum result in an improvement of 1.7 Gy (for bladder without any smoothing, p=0.003). The approximately horizontal lines in Fig. 1b show that the extent of smoothing did only marginally impact dosimetric plan quality. Although spacing, smoothing and geometrical steering had only a minor impact on population-mean dosimetric parameters for the MAIN structures, there were large inter-patient variations (see Fig.2 for impact on bladder D2cc). When using 1mm spacing instead of 5mm, reduction in bladder D2cc varied between -9.2Gy and +0.5 Gy, depending on the patient (Fig. 2a). Fig.2b shows large inter- patient variations in the impact of smoothing. Conclusion An automated multi-criteria planning workflow for EMBRACE II adaptive BT for LACC was used to investigate the impact of imposed loading pattern requirements on dosimetrical plan quality. Population-mean dosimetric plan parameters did only marginally improve when these requirements were relaxed. However, large inter-patient variations were observed, pointing at options to improve treatment quality with patient-specific choices for loading pattern requirements.