Normal Tissue Complication Probability Model for Acute Radiation Dermatitis in Patients With Head and Neck Cancer Treated With Carbon Ion Radiation Therapy

Purpose: This study aimed to explore the prognostic factors associated with acute radiation dermatitis (ARD). A normal tissue complication probability (NTCP) model for ARD in patients with head and neck cancer treated with carbon ion radiation therapy was developed. Materials and methods: A total of 187 patients were included in the analysis, and the endpoint was >= grade 2 ARD. The biological and physical dose-surface parameters associated with ARD were used in the logistic regression model. The mean areas under the receiver operating characteristic curve in the internal cross-validation and Akaike's corrected Information Criterion were examined for model evaluation and selection. The multivariate logistic regression NTCP models were established based on factors with weak correlation. Results: Tumor volume, planning target volume to the skin, radiation technique and all dose-surface parameters were significantly associated with ARD (P <.05). Models with high performance for grade 2 to 3 ARD were constructed. The most significant prognostic predictors were S-40 Gy (relative biological effectiveness,S- RBE) and S-20 Gy (absolute surface area receiving RBE-weighted dose of 40 Gy (RBE) or physical dose of 20 Gy). The internal cross-validation-based areas under the receiver operating characteristic curve for models with S-40 Gy (RBE) and S-20 Gy were 0.78 and 0.77, respectively. The biological and physical dose-surface parameters had similar performance at various dose levels. However, the performance of the multivariate NTCP models based on 2 factors was not better than that of the univariate models. Conclusions: NTCP models for ARD may provide a basis for the development of individualized treatment strategies and reduce the incidence of severe ARD in patients with head and neck cancer receiving carbon ion radiation therapy. Furthermore, biological and physical dose-surface parameter-based models are comparable. However, further validation with more evaluation parameters is warranted. (C) 2022 Elsevier Inc. All rights reserved.