ESTRO 2025 - Abstract Book

S122

Invited Speaker

ESTRO 2025

4850

Speaker Abstracts Radiation omission trials: The dust has settled - How to revisit and optimise radiation oncology clinical trial design in smarter ways? Nina Sanford Radiation Oncology, University of Texas Southwestern, Dallas, USA Abstract: Omission trials are important in oncology because they assess whether therapy can be safely de-escalated, thereby improving quality of life without compromising cancer outcomes. As compared to other sub-specialties, there are proportionally more omission trials in radiation oncology. In this session, we will first review the rationale, different types and statistics behind therapy omission trials and contextualize the overall radiation omission trial landscape. Next, we will discuss several recently published, high profile trials assessing radiation omission or de-escalation across disease sites, focusing on methodology and interpretation. Lastly, we will propose suggestions for optimizing the design and implementation of radiotherapy omission trials going forward – including considerations regarding endpoints, control arms, and non-inferiority margins, emphasizing the importance of fostering collaboration between oncologic specialties. Speaker Abstracts Autoplanning: Supported by clinical evidence and motivated by clinical need Linda Rossi Radiotherapy, Erasmus MC, Rotterdam, Netherlands Abstract: Radiotherapy treatment plan preparation is a challenging and time-consuming process that requires advance skills, expertise and sufficient time. With Artificial Intelligence reaching now various field of our lives, we have been pioneers in radiotherapy in understanding the potential for our clinics. Automation of treatment planning have been luckily highly investigated and developed in the last decade(s). In this lecture, we will i) examine the most common techniques for automated planning in radiotherapy, ii) discuss the advantages that automated planning has provided, and iii) identify the areas where automated planning can further and still support our clinical needs. More in details, basis of AI-data driven and AI-rule driven automated planning solution will be shown, as knowledge based planning, multi-criteria planning, protocol-based automatic iterative optimization and deep learning dose prediction. Principle mathematical background of the different solutions will be seen as well an characteristic of clinical implementation. While all methodologies aim to streamline planning, reduce workload, and enhance quality, each of them present different advantages and disadvantages. High amount of publications supports autoplanning, showing that automated systems can produce plans comparable or superior to manual ones while significantly reduces planning time. Moreover improvement in consistency is the other big benefit brought from automated planning. The need for autoplanning arises from complex treatment sites and combined treatment, making workflow more articulate. Next to it, there is the high demand for personalized care and the development of real-time adaptive planning with more complex workflow and with time pressure. This all under a constant increase in tumor incidence and decline in worker availability in clinical care. Automated planning serves to automate this process, with 4852