ESTRO 2023 - Abstract Book

S476

Sunday 14 May 2023

ESTRO 2023

Figure 1. Summary of participants' points of view on using AI-based decision aids in different scenarios Conclusion This is the first qualitative study exploring patient views on the use of AI-based decision aids in oncology. It highlights that patients are willing to use AI-based tools to assist in medical decision-making, but are wary of using them in high-impact decisions such as survival and recurrence. The role of the medical doctor in these scenarios is also substantial. It is therefore crucial to involve patients in the development process of AI and decision aids in order to adopt technology successfully. Furthermore, introducing AI systems as a source of information rather than a substitute for the physician is an effective strategy for encouraging patients to embrace new technologies. OC-0595 Proton FLASH: Impact of beam pauses on the biological response in an acute damage mouse model B.S. Sørensen 1 , M.K. Sitarz 1 , C. Ankjærgaard 2 , L. Kristensen 3 , C.E. Andersen 4 , C. Grau 1 , P. Poulsen 1 1 Aarhus University Hospital, Danish Centre for Particle Therapy, Aarhus, Denmark; 2 DTU , Health Tech, Roskilde, Denmark; 3 Aarhus University Hospital, Department of Experimental Clinical Oncology, Aarhus, Denmark; 4 DTU, Health Tech, Roskilde, Denmark Purpose or Objective Preclinical studies indicate a normal tissue sparing effect using ultra-high dose rate (FLASH) radiation with comparable tumor response. This differential response is promising in order to obtain improved clinical outcome. We have previously conducted an in vivo study validating the effect of PBS proton FLASH on acute skin toxicity, and demonstrated that a 44- 58% higher dose was needed to obtain the same biological response when using proton FLASH. The aim of the present study was to use the same model to investigate the biological effect of introducing short pauses in the dose delivery, emulating a clinical situation with FLASH delivery with multiple treatment fields. Materials and Methods The right hind limbs of non-anaesthetized CDF1 mice were irradiated in the entrance plateau of a scanning proton pencil beam. The delivered fields were 2cm x 3cm to ensure a homogeneous dose across the whole mouse leg. The irradiation dose was 39.3Gy using field dose rates of 60Gy/s for FLASH and 0.4Gy/s for conventional delivery (CONV). The FLASH dose was given either as a single delivery or split into 2, 3, 4 or 6 identical deliveries with 2 minutes pauses between each delivery. For example, splitting into 2 deliveries meant 19.65Gy followed by a pause of 2 minutes followed by 19.65Gy with the same field (100% overlap of the fields). The CONV dose was given in either 1 or 6 deliveries with 2 minutes pauses. 9- 11 mice were included in each treatment group. The endpoints were different levels of skin toxicity within 25 days post irradiation. Statistical significance between FLASH and CONV groups (for 1 and 6 deliveries) were calculated with Fisher´s exact test. Results For CONV, all mice displayed skin toxicity at all levels when irradiated with 39.9Gy in one delivery, and that was unaltered by splitting the dose into 6 deliveries (Fig 1). For FLASH, delivering the dose unsplit revealed comparable outcome to previously published data with no animals displaying moderate or severe damage (score 2.5-3, Fig 1). The reduced toxicity for FLASH with 1 delivery relative to CONV was significant for score 2.0 (p <0.001), score 2.5 (p <0.0001) and score 3.0 (p <0.0001). Splitting the FLASH dose into 2 deliveries reduced the FLASH effect, but the normal tissue sparing effect was still observed comparing to CONV. For 3 or more deliveries, the FLASH effect was almost abolished for lower grades of acute toxicity. With 6 deliveries, the toxicity for FLASH relative to CONV was not significant for any of the scores (p>2 – p>0.99). Proffered Papers: Normal tissue radiobiology