ESTRO 2023 - Abstract Book

S845

Tuesday 16 May 2023

ESTRO 2023

1 Aarhus University Hospital, Danish Center for Particle Therapy, Aarhus, Denmark

Abstract Text FLASH is radiotherapy delivered with ultra-high dose rates that are hundreds of times higher than in conventional radiotherapy. It has received massive interest since pre-clinical experiments have shown a preferential sparing of healthy tissue with maintained effects on tumors. Most experiments have used electron beams with shallow treatment depths of a few centimeters. FLASH dose rates can also be obtained with high energy proton beams at clinical proton facilities, which allows treatment in any relevant tumor depth. In this lecture, a brief overview of proton FLASH will be given including realization of FLASH dose rates with proton beams, proton FLASH dosimetry, treatment planning and pre-clinical studies at the Danish Center for Particle Therapy.

Symposium: Handling longitudinal imaging data

SP-1046 Temporal volume changes during online adaptive MRgRT: Potential and challenges U. Bernchou 1 1 Odense University Hospital, Department of Oncology, Odense, Denmark

Abstract Text Many anatomical and pathological volumetric changes occur on time scales relevant to a radiotherapy treatment course. Examples include tumour growth and regression, organ filling, peristalsis, movement of gas pockets, and respiration. In a traditional CT-based radiotherapy workflow, some of these volumetric changes may be partly counteracted by the use of plan libraries or auxiliary imaging needed for offline re-planning. However, in most cases, large PTV margins are needed to ensure target coverage, although this may lead to excess dose to the OAR. Furthermore, large PRV margins are often needed to ensure a safe and tolerable treatment, which again may lead to insufficient target coverage. Online adaptive MRgRT has the potential to address many of the issues coursed by volumetric changes. During the MRgRT workflow, target and OAR structures may be propagated from the planning images to the 3D MR images acquired on the MR-linac using deformable image registration. Furthermore, the superior soft-tissue image quality of MR allows for accurate adjustment of the target and OAR delineation while the patient is on the treatment couch. Thereby, the treatment plan may be adapted to the anatomical situation of the day. Additionally, 2D cine MR imaging acquired during treatment delivery may be used to monitor target drift or respiration-induced motion. Some MR-linac systems can account for such movement using couch- or beam-shifts and beam-gating. Therefore, online adaptive MRgRT can lead to reduced PTV and PRV margins since the exact location of targets and OAR is known at the time of treatment. Although online adaptive MRgRT holds great potential for many disease sites, the workflow is labour-intensive and time consuming. Dedicated oncologists and medical physicists must often be present during each treatment fraction. Furthermore, 3D MR imaging and manual contour adjustment are relatively slow processes, which may lead to volumetric organ changes during the adaptation process. Such changes cannot always be accounted for using beam-gating and couch- or beam-shifts. During this talk, a short introduction to the MRgRT workflow will be presented. Examples of volumetric changes occurring during a single treatment fraction as well as the whole treatment course will be given based on recent literature and clinical experience. The potential benefits and challenges of these volumetric changes will be discussed. Strategies for speeding up the workflow or otherwise reducing the effect of intra-fractional volume changes will be presented. SP-1047 Handling imaging data for future adapt-to-response workflows J. Habrich 1 1 Section for Biomedical Physics, Department of Radiation Oncology, University of Tübingen, Tübingen, Germany Abstract Text Over the course of a conventional radiotherapy treatment, additional magnetic resonance images or positron emission tomography (PET) images require auxiliary time and effort, often across different clinical departments, like radiology or nuclear medicine. Therefore, an adaptation of a treatment plan during therapy is challenging and may only be realized unfrequently or at specific pre-defined timepoints. Hybrid MR-Linacs, combining a linear accelerator with magnetic resonance imaging, address this problem by providing the possibility of daily magnetic resonance imaging (MRI) during fractionated radiotherapy. Hence, the position of the target structures as well as organs at risk can be visualized on a daily basis and the positioning of the patient as well as the daily treatment plan can be adapted online. By doing so, margins around the target volumes and doses to organs at risk can be minimized. While anatomical imaging at the start or during a treatment fraction may improve the accuracy of the radiation delivery, little information about the individual response to therapy is gained. Therefore, various recent studies evaluated quantitative imaging biomarkers (QIBs) for their potential with regard to correlating QIB to radiotherapy outcome and identifying predictive or prognostic QIBs representing the tumor response to radiotherapy. Usually, these QIBs are derived from functional MRI or PET, which in the case of MRI can also be acquired daily on the MR-Linac. Possible biomarkers are the apparent diffusion coefficient (ADC) from diffusion weighted MRI, the transfer constant (Ktrans) from dynamic contrast enhanced imaging or the standard uptake value from PET. But the integration of QIBs in the standard clinical workflow in the context of therapy adaptation based on a QIB has yet to be done. Apart from validating the acquisition of a QIB and its prognostic value, the technical integration of the images into the clinical workflow is quite difficult, especially in online adaptive radiotherapy on MR-Linacs. This talk will discuss different strategies on how to implement online as well as offline adapt-to-response workflows, firstly. Different QIB have shown predictive or prognostic potential, including changes of tumor volume. Furthermore, the necessary steps for the translation of a biomarker into a clinical workflow will be presented, including determining the QIB, technical validation of the acquisition and correlation of the QIB to outcome parameters. The longitudinal assessment of diffusion-weighted MRI as a response marker for head and neck cancer radiotherapy will be described and results on diffusion-weighted MRI in head and neck cancer assessed on a 1.5 T MR-Linac including repeatability and reproducibility will be presented. Finally, an example for a response adaptive workflow out of a phase I trial on diffusion-weighted MRI