Control of dental calculus Prevents severe Radiation-Induced oral mucositis in patients undergoing radiotherapy for nasopharyngeal carcinoma.

PURPOSE

This study aims to develop an artificial intelligence model to predict severe radiation-induced oral mucositis (RIOM) in patients with locally advanced nasopharyngeal carcinoma (LA-NPC) and verify the risk factors associated with severe RIOM.

METHODS AND MATERIALS

A total of 578 patients diagnosed with LA-NPC and undergoing radiotherapy were enrolled in this study. This cohort comprised 430 retrospective patients used for model development/validation, and 148 patients for the prospective verification study. Multifaceted data related to RIOM were collected to build an explainable multi-classifier fusion (MCF) model to identify severe RIOM associated risk factors. A prospective study was designed to validate the key risk factors.

RESULTS

The MCF model demonstrated satisfactory performance in severe RIOM prediction when integrating all dosimetric, clinical, and oral features, with an AUC of 0.904, ACC of 0.849, SEN of 0.853 and SPE of 0.846 on the independent testing set. The dental calculus index of 2 was identified as a significant key risk factor for developing RIOM. The severe RIOM rate in the prospective intervention cohort was 8.1 % (95 % CI:4.3 %∼13.7 %), lower than that in the model development cohort, with a decrease of 31 % (95 % CI23.9 %∼36.8 %, p < 0.0001).

CONCLUSIONS

The developed model can serve as a valuable tool for providing timely alerts for high-risk patients with the severe RIOM and assisting physicians in optimizing treatment management. The dental calculus index is a key independent risk factor for severe RIOM. The effective control of the dental calculus can significantly mitigate the onset of severe RIOM.

CLINICALTRIALS

gov: NCT05858385.