ESTRO 2024 - Abstract Book

S1272

Clinical - Head & neck

ESTRO 2024

The multi-endpoint model had a good predictive performance in terms of discrimination and could be used for identifying high risk patients for future clinical trials of for example dose escalation. UICC8 staging, however, performed equally well and would provide a simpler inclusion criterion, despite being developed with overall survival as endpoint.

Keywords: risk modelling, validation, UICC8

References:

1. Håkansson K, Rasmussen JH, Rasmussen GB, et al. A failure-type specific risk prediction tool for selection of head and-neck cancer patients for experimental treatments. Oral Oncol. 2017;74:77 – 82.

2. Gerds, Thomas A; Scheike, Thomas H; Andersen PA. Absolute Risk Regression For Competing Risks: Interpratation, Link Functions and Prediction. Stat Med. 2012;31(29):3921 – 30.

3. Blanche P, Dartigues J-F, Jacqmin-Gadda H. Estimating and comparing time-dependent areas under receiver operating characteristic curves for censored event times with competing risks. Stat Med. 2013;32(30):5381 – 97.

1236

Mini-Oral

Towards the global integration of a comprehensive toxicity risk profile for HNC patients

Suzanne P.M. de Vette 1 , Maria I. van Rijn-Dekker 1 , Lisa Van den Bosch 1 , Kylie Keijzer 2 , Hendrike Neh 1 , Hung Chu 1 , Yan Li 1 , Mark L. Frederiks 1 , Hans Paul van der Laan 1 , Jolien Heukelom 3 , Arjen van der Schaaf 1 , Katherine A. Hutcheson 4 , Roel J.H.M. Steenbakkers 1 , Nanna M. Sijtsema 1 , Clifton D. Fuller 5 , Johannes A. Langendijk 1 , Amy C. Moreno 5 , Lisanne V. van Dijk 1 1 UMCG, Radiotherapy, Groningen, Netherlands. 2 UMCG, Hematology, Groningen, Netherlands. 3 Maastro Clinic, Radiotherapy, Maastricht, Netherlands. 4 MDACC, Head and Neck Surgery, Houston, USA. 5 MDACC, Radiation Oncology, Houston, USA

Purpose/Objective:

Normal tissue complication probability (NTCP) models predict the development of toxicity after radiotherapy, and are currently used in routine clinical practice to guide treatment planning and/or to select the optimal treatment modality to minimize toxicity. The comprehensive individual toxicity risk (CITOR) project by Van den Bosch et al. (2021) developed NTCP models for 22 most common toxicities for head and neck cancer (HNC) patients, divided into six domains: swallowing, salivary, mucosal, speech, pain and general. These models were developed in a single Dutch institute and subsequently validated in other Dutch centres. To further demonstrate the potential of these models and to promote global integration, this study aims to evaluate the performance of the CITOR NTCP models