ESTRO 2024 - Abstract Book

S5155

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2024

Severe hypoxia were strongly related to the parameters representing the metastatic cascade; pT, LVI and LN (p<0.0001). Interestingly, Gleason Score and Longest axis were most strongly associated with moderate hypoxia (p<0.001). Severe hypoxia was the hypoxia level that was most strongly associated with PSA relapse (p<0.0001). Importantly, the presence of severe hypoxia was shown to provide independent prognostic information in multivariate analysis including the current treatment decision variable D’Amico risk classification, Gleason Score and Tumor stage.

Conclusion:

CSH imaging can quantify different hypoxia levels. Severe hypoxia is strongly associated with steps in the metastatic cascade and a poor outcome to prostatectomy. CSH imaging may be useful for selecting patients in need of lymph node irradiation.

Keywords: hypoxia, MR, metastasis

References:

1) Hompland et al. Cancer Res. 2018

3126

Poster Discussion

Strategies for continuous NTCP model adaptation in an evolving esophageal cancer treatment landscape

Gilles Defraene 1 , Pieter Populaire 1,2 , Karin Haustermans 1,2

1 KU Leuven, Laboratory of Experimental Radiotherapy, Leuven, Belgium. 2 UZ Leuven, Department of Radiotherapy, Leuven, Belgium

Purpose/Objective:

The rapid evolution of state-of-the-art cancer treatments leads to a systematic underperformance of prediction models of toxicity, i.e., normal tissue complication probability (NTCP) models, in recently treated patients. Smoothly adapting NTCP models to relevant treatment changes is required, e.g. to ensure optimal individualized treatment planning or model-based patient selection for proton therapy at all times. It is unclear how to best adapt NTCP models when confronted with continuously incoming patient data. The reaction to changes should not be too slow (as in classical retrospective model validation) nor too fast (to avoid fitting associations that later appear irrelevant). Different strategies could therefore be envisioned going from the use of all available retrospective data in order to obtain a universal model, to the use of only a recent subset of patients for a model specific for a new treatment era. We evaluated the impact of 3 NTCP model adaptation strategies.