ESTRO 2025 - Abstract Book

S1081

Clinical – Head & neck

ESTRO 2025

3402

Digital Poster Integrating Dosimetric and Tooth Avulsion Status for Personalized Risk Prediction of Mandibular

Osteoradionecrosis in Head and Neck Cancer Ahmed Allam Mohamed, Julian Leeder, Michael Eble Radiation Oncology, Faculty of Medicine, RWTH Aachen University, Aachen, Germany

Purpose/Objective: Mandibular osteoradionecrosis (ORN) is a serious complication in head and neck cancer (HNC) patients treated with radiotherapy (RT), with an incidence of 3–15%. It arises due to radiation-induced bone damage, often necessitating surgical intervention. This study aimed to identify clinical and dosimetric factors associated with ORN risk and develop a predictive model to support personalized preventive strategies. Material/Methods: A retrospective case-control study was conducted on 298 head and neck cancer (HNC) patients who underwent RT or chemoradiotherapy between January 2012 and May 2020. Eligible patients had a follow-up period of over six months and received a radiation dose of ≥50 Gy. Clinical data collected included age, tumor site, smoking history, and tooth avulsion status. Dosimetric parameters analyzed comprised mandibular volume, V10–V60, mean dose (Dmean), dose near maximum (D2%), and maximum dose (Dmax). A correlation matrix was utilized to identify the parameters most strongly associated with ORN events, and logistic regression models were applied to assess the predictive impact of key variables. Results: The median follow-up duration for the cohort was 32.43 months. Mandibular ORN was observed in 20 patients (6.7%), corresponding to a 5-year cumulative incidence of 7.4%. The correlation matrix identified post-treatment tooth avulsion and V40 as having the strongest correlation to ORN events (Figure 1). Logistic regression analysis confirmed V40 as a significant predictor, with a coefficient of 0.0471 (p = 0.004). The model demonstrated modest explanatory power, with a pseudo-R-squared value of 0.0682, and its overall significance was supported by a log-likelihood ratio (LLR) test (p = 0.0016) (Figure 2). In the adjusted logistic regression model that included both V40 and post-treatment tooth avulsion, V40 remained significant, with a coefficient of 0.0545 (p = 0.003). post-treatment tooth avulsion was also strongly associated with ORN events, with a coefficient of 1.3850 (p < 0.001). The inclusion substantially improved the model's performance, increasing the pseudo-R-squared value to 0.2471 and significantly enhancing the model's fit compared to the V40-only model (Figure 2).