Risk factors and prediction models for severe radiation-induced oral mucositis in patients with nasopharyngeal carcinoma undergoing chemoradiotherapy.

OBJECTIVE

This study aimed to investigate the factors associated with severe radiation-induced oral mucositis (SRIOM) in nasopharyngeal carcinoma (NPC) patients undergoing chemoradiotherapy (CRT) and to establish a prediction model for SRIOM.

METHODS

A total of 262 NPC patients who underwent CRT were analyzed retrospectively, including 192 in the modeling group and 70 in the validation group. The modeling group was divided into the non-SRIOM group (n = 112) and the SRIOM group (n = 80), and the validation group was divided into the non-SRIOM group (n = 40) and the SRIOM group (n = 30) according to the presence of SRIOM. Univariate and multivariate logistic logistic analyses were performed on the clinical data and general characteristics of all patients to construct a prediction model for SRIOM in NPC patients. The practical efficacy of the prediction model was evaluated using Hosmer-Lemeshow test, receiver operating characteristic curve (ROC), and decision curve analysis (DCA).

RESULTS

BMI < 23.9 kg/m2, history of periodontal disease, history of alcohol consumption, history of smoking, non-use of oral mucosal protectants, and poor oral hygiene were independent risk factors for SRIOM in NPC patients. The prediction model showed an area under the ROC curve of 0.813 (95% CI 0.752-0.875). The prediction model demonstrated strong predictive accuracy and clinical utility, as evidenced by both calibration and DCA curves.

CONCLUSION

The SRIOM prediction model, developed from the clinical characteristics and general information of NPC patients, is beneficial in clinical practice for identifying high-risk SRIOM and creating tailored treatment plans.