1-4 December 2024, Vienna, Austria

Course Report

My name is Ragu, and I am a clinical oncology registrar based in London, UK. Currently, I am a clinical research fellow undertaking a PhD at the Institute of Cancer Research and the Royal Marsden Hospital.

This was my first attendance at an ESTRO course. It was highly recommended by colleagues who had previously attended. Given that my PhD involves dose-toxicity modelling and work with clinical data, the course content was particularly relevant to my ongoing research projects and academic interests.

The most interesting aspects for me were the sessions on dose-response and outcome modelling. These topics are immediately applicable to my research. The lecturers presented complex concepts in an easy-to-follow manner, particularly the daunting mathematical equations.

The course significantly improved my understanding of dose-response modelling, methods by which to assess the performance and validity of models, and, importantly, the limitations I should be aware of. It provided a good overview of statistical methods that are commonly used in clinical research, such as regression analysis, survival statistics, and sample size calculations. As a result, I feel better equipped to appraise research studies critically and more confident to perform these analyses for my research.

The course met my expectations. I gained a deeper understanding of radiation-dose-response and toxicity models and saw examples of their utility in clinical practice. For instance, we were shown how normal tissue complication probability (NTCP) models could be used to select patients who were likely to benefit from proton therapy. Importantly, I learned how to assess model validity, understand challenges and considerations before integrating models into clinical practice, and recognise their key limitations.

Three important ‘takeaways’ from the course were:

• before developing models, it is essential to plot and visually examine data associations to identify trends or patterns;
• traditional NTCP modelling that uses single dose-volume histogram parameters loses spatial information, but alternatives such as convolutional neural networks, voxel-based analysis and image-response models may address these limitations; and
• beyond traditional randomised controlled trials, innovative designs such as basket, umbrella, and platform studies offer promising approaches for clinical research.

 

The knowledge gained will directly benefit my current research role, as I am now more aware of ways to validate models and understand their limitations. The course also provided an overview of alternative analysis methods that can be used in dose-toxicity modelling that incorporate spatial information, which could be an avenue for further research. Additionally, the lectures on statistics and clinical trials have improved my ability to critique the statistical methodologies that are used in clinical studies, which is a valuable skill as a clinician.

If you are interested in radiotherapy dose-toxicity/response modelling or radiotherapy clinical trials, this course provides an excellent overview of essential topics. It is both highly educational and practical, making it a valuable resource for anyone with these interests. I would definitely recommend it to colleagues with similar research interests.

Dr Ragu Ratnakumaran

Clinical research fellow

Royal Marsden Hospital and Institute of Cancer Research

London, UK

ragu.ratnakumaran@icr.ac.uk