S2998
Invited Speaker
ESTRO 2026
clinical workflow is challenging in a single institution and even worse when moving to multiple institutions and/or national datasets. In Denmark, the Danish Breast Cancer Group (DBCG) initiated a national retrospective data collection of high-risk breast cancer patients treated with radiotherapy in the period 2008 to 2016, the DBCG RT Nation cohort[1,2]. From the DBCG database, 9100 patients were identified to be treated in 2008 to 2016. During data collection, curation and standardization, more than 1500 patients were excluded due to data availability and incomplete or inconclusive data. The final database contained 7448 breast cancer patients, including planning CT scans, structure sets, treatment plans and dose distributions. The data was collected from seven different radiotherapy centres and from three different planning systems. To handle the dataset, systematic harmonization was applied to address the inter-centre variability. Despite the challenges of collecting large national databases, their value is considerable as they enable a wide range of research applications. Retrospective, population-based datasets can be used for dose audits, describing the planned treatments delivered to patients at a national level, as well as analyses of adherence to clinical guidelines. When linked with other databases, such as registers of cause of death, the dataset further enables investigation of dose- response relationships. In addition, large-scale databases provide a good foundation for the development and validation of AI-based tools. In this talk, the lessons learned from collecting the RT Nation cohort will be discussed, spanning from data extraction and harmonization to quality assurance. References: [1] Refsgaard L, Skarsø ER, Ravkilde T, Nissen HD, Olsen M, Boye K, et al. End-to-end framework for automated collection of large multicentre radiotherapy datasets demonstrated in a Danish Breast Cancer Group cohort. Physics and Imaging in Radiation Oncology 2023;27:100485. https://doi.org/10.1016/j.phro.2023.100485. [2] Refsgaard L, Buhl ES, Yates E, Maae E, Berg M, Al-Rawi S, et al. Evaluating Danish Breast Cancer Group locoregional radiotherapy guideline adherence in clinical treatment data 2008–2016: The DBCG RT Nation study. Radiotherapy and Oncology 2024;199. https://doi.org/10.1016/j.radonc.2024.110289.
national cancer registries, most of these specific to a cancer type. These contain patient-specific information about diagnosis, treatment and outcome. Large retrospective datasets can be created by combining information from SKvaRT and cancer-specific registries. However, national registries typically include limited or no data on radiotherapy side effects; the responsibility for monitoring side effects is largely left to individual clinics. Few clinics have the resources or the infrastructure for systematic monitoring, resulting in, at best, limited retrospective analyses of clinical records. Karolinska University Hospital, treating 25% of the Swedish head and neck cancer radiotherapy patients, implemented a local quality registry in 2014, with a focus on late side effects. It contains over 3000 patients, and a dedicated research program is linked to the registry. It has been used to evaluate reirradiation and dose escalation in head and neck cancer, with a current focus on prediction model development and validation in real-world cohorts, for evidence-based treatment planning. In the EU-funded TETRIS project, the Swedish breast cancer registry and other national quality registries have been queried to, in combination with treatment data, create a large cohort for testing published prediction models. While the registries do not contain side effect variables, surrogate variables like prescribed medication can be used, as well as conditions recorded as new diagnosis codes, like heart disease or second cancers. In our experience, effective use of registries for research requires expertise regarding the recorded variables and their respective quality. However, with the establishment of SKvaRT, the power of national quality registries can now be harnessed also for radiotherapy research. References: https://cancercentrum.se/diagnosbehandling/stralbeh andling/kvalitetsregisterforstralterapi.7385.html https://projectetris.eu/ 5310 Lessons learned from a nation-wide radiotherapy database: the Danish experience Emma Skarsø Buhl Department of Clinical Medicine, Aarhus University, Aarhus, Denmark. Danish Center for Particle Therapy, Aarhus University Hospital, Aarhus, Denmark Radiotherapy clinics worldwide generate large amounts of clinical and imaging data daily, however utilization of these data for research and clinical development remains limited. Clinical data outside of trials offers information that can be used to increase the understanding of treatment effectiveness. However, collecting and organizing datasets from the
5312 Physics highlights and innovations Åsa Carlsson Tedgren
Nuclear Medicine and Medical Physics, Karolinska University hospital, Stockholm, Sweden. Oncology Pathology, Karolinska Institute, Stockholm, Sweden
Made with FlippingBook - Share PDF online