ESTRO 2024 - Abstract Book

S111 ESTRO 2024 I will focus on face validity, domain knowledge and the power of data visualisation as a key components in data analysis to be prioritised as high as p-value calculations. No teaching session on statistics without discussion of the reproducibility crisis of science, however. We will discuss case examples from recent works of typical questions related to pre-planning of analyses. I will also discuss how to improve the trust in more advanced analyses including machine learning methods. Invited Speaker

It should be clear at the end of the talk that the "numerical literacy" of a medical physicist combined with the domain knowledge are a very good starting point for good science.

3505

Surgery: When and how?

Magnus Nilsson

Karolinska Institutet, CLINTEC, Stockholm, Sweden

Abstract:

Currently the standard of care curative intent treatment for oesophageal cancer is multimodal, including surgery for adenocarcinoma, while for squamous cell carcinoma it can either be neoadjuvant chemoradiotherapy (or in Asia chemotherapy alone) followed by planned surgery, or definitive chemoradiotherapy with surveillance and salvage surgery only when needed to achieve local tumour control in non-complete responders or patients with local recurrence. Two randomised trials have addressed the issue if surgery can be left out completely in clinical responders to chemoradiotherapy but failed to convince the medical community of this so far. This is partly because of alleged issues with the trial design, but also because a very high proportion of the clinical responders turned out to progress locally or have remaining tumour after being operated. These trials clearly demonstrate that clinical response evaluation after chemoradiotherapy is difficult and the negative predictive value of clinical response evaluation regarding remaining tumour is usually low. In the recently reported SANO trial another but similar approach was used. Patients who had completed neoadjuvant CROSS-type chemoradiotherapy underwent clinical response evaluation at 6 and 12 weeks and were if still clinically without histological proof of remaining tumour, randomised to either surgery or surveillance. In the preliminary data reported so far surveillance seems to be non-inferior regarding overall survival, but conclusions cannot be drawn until the full data have been published, including stratification by histological type and clarification of the intention to treat population. In addition, generalizability can be questioned as the majority of patients who were clinical non-complete responders, and operated outside the trial, may have suffered detriment from the long delay of surgery of >10 weeks. In my view surgery is likely to remain a part of the primary curative treatment for fit patients with oesophageal adenocarcinoma until a game-changing new systemic therapy may eventually emerge in the future. On the other hand, I am convinced that for oesophageal squamous cell carcinoma, which is still globally the vastly more common of the two subtypes, a primarily non-surgical approach with definitive chemoradiotherapy followed by surveillance and salvage surgery only when really needed could become the only standard of care in the not too distant future, with non-inferior overall survival and superior quality of life, due to the avoidance of surgery in a majority of patients. This issue is currently being tested in the global randomised NEEDS trial.