Prediction of radiation-induced mucositis of H&N cancer patients based on a large patient cohort.

PURPOSE/OBJECTIVE: Radiation-induced mucositis is a severe acute side effect, which can jeopardize treatment compliance and cause weight loss during treatment. The study aimed to develop robust models to predict the risk of severe mucositis. MATERIALS/METHODS: Mucosal toxicity scores were prospectively recorded for 802 consecutive Head and Neck (H&N) cancer patients and dichotomised into non-severe event (grade 0-2) and severe event (grade 3+) groups. Two different model approaches were utilised to evaluate the robustness of the models. These used LASSO and Best Subset selection combined with 10-fold cross-validation performed on two-thirds of the patient cohort using principal component analysis of DVHs. The remaining one-third of the patients were used for validation. Model performance was tested through calibration plot and model performance metrics. RESULTS: The main predicted risk factors were treatment acceleration and the first two principal dose components, which reflect the mean dose and the balance between high and low doses to the oral cavity. For the LASSO model, gender and current smoker status were also included in the model. The AUC values of the two models on the validation cohort were 0.797 (95%CI: 0.741-0.857) and 0.808 (95%CI: 0.749-0.859), respectively. The two models predicted very similar risk values with an internal Pearson coefficient of 0.954, indicating their robustness. CONCLUSIONS: Robust prediction models of the risk of severe mucositis have been developed based on information from the entire dose distribution for a large cohort of patients consisting of all patients treated H&N for within our institution over a five year period.

NTCP model validation method for DAHANCA patient selection of protons versus photons in head and neck cancer radiotherapy.

Introduction: Prediction models using logistic regression may perform poorly in external patient cohorts. However, there is a need to standardize and validate models for clinical use. The purpose of this project was to describe a method for validation of external NTCP models used for patient selection in the randomized trial of protons versus photons in head and neck cancer radiotherapy, DAHANCA 35. Material and methods: Organs at risk of 588 patients treated primarily with IMRT in the randomized controlled DAHANCA19 trial were retrospectively contoured according to recent international recommendations. Dose metrics were extracted using MatLab and all clinical parameters were retrieved from the DAHANCA database. The model proposed by Christianen et al. to predict physician-rated dysphagia was validated through the closed testing, where change of the model intercept, slope and individual beta's were tested for significant prediction improvements. Results: Six months prevalence of dysphagia in the validation cohort was 33%. The closed testing procedure for physician-rated dysphagia showed that the Christianen et al. model needed an intercept refitting for the best match for the Danish patients. The intercept update increased the risk of dysphagia for the validation cohort by 7.9 ± 2.5% point. For the raw model performance, the Brier score (mean squared residual) was 0.467, which improved significantly with a new intercept to 0.415. Conclusions: The previously published Dutch dysphagia model needed an intercept update to match the Danish patient cohort. The implementation of a closed testing procedure on the current validation cohort allows quick and efficient validation of external NTCP models for patient selection in the future.