Combining clinical and dosimetric features in a PBS proton therapy cohort to develop a NTCP model for radiation-induced optic neuropathy

Purpose Radiation-induced optic neuropathy (RION) is a rare, yet severe complication following radiation therapy for brain, head and neck, or skull-base tumors. Although several risk factors, such as age, metabolic syndrome, and delivered dose, have been identified, we aimed at expanding the understanding of the mechanisms of interplay regarding dosimetry and patient variables leading to the onset of RION with a focus on proton therapy. Methods and Materials In this retrospective study, we have investigated proton-specific risk factors by comparing common phenomenological normal tissue complication probability models with a multivariate analysis that includes clinical features on a cohort of patients with skull-base and head and neck cancer treated with pencil beam scanning. Results Although predictive power of the Lyman-Kutcher-Burman and Poisson models was limited for this data set, the addition of clinical variables such as age, tumor involvement, hypertension, or sex remarkably increased model performance. Conclusions Based on our assessment, the maximum dose in the optical apparatus is confirmed the most intuitive risk factor. However, above a certain dose threshold, clinical patient characteristics are the deciding factors for the onset of RION. We observed a tendency toward a volume effect that, if confirmed, would imply a benefit for high precision radiation therapy techniques such as proton therapy for the treatment of patients with high clinical risk for RION.