Risk-based treatment optimisation to reduce radiation-induced toxicity in head and neck cancer patients

Radiotherapy plays an important role in the treatment of head and neck cancer (HNC) patients. However, the tumour cannot be adequately radiated without irradiating healthy normal tissue as well. Irradiation of healthy tissues results in numerous toxicities (e.g., xerostomia (dry mouth syndrome) or severe swallowing problems) that significantly impair their quality of life. New radiotherapy treatment techniques, like protons, enable a more precise irradiation of the tumour, thereby reducing the radiation dose to these healthy tissues. The main challenge is how to arrive at the 3D dose distribution that cures patients with the lowest toxicity burden. This process requires detailed information on the relationship between the 3D dose distribution and the risk of toxicities. This relationship is described in multivariable prediction models, commonly referred to as Normal Tissue Complication Probability (NTCP) models. In this thesis we developed and validated a comprehensive set of NTCP models that predict the risk of various common radiation-induced toxicities. Together these NTCP models converge into a corresponding risk profile, the Comprehensive Individual Toxicity Risk (CITOR) profile. It provides new insight into the involvement of healthy tissues in the development of toxicities, indicating which healthy tissues are more important to spare. The oral cavity appears to be the predominant healthy tissue that is associated with multiple toxicities and should therefore be spared with a high priority. Some of the NTCP models of this CITOR profile are currently used nationally in clinical practice to optimise the dose distribution and to select HNC patients for proton therapy.