Risk factor analysis and development of predictive models for osteoradionecrosis in patients with nasopharyngeal carcinoma after concurrent chemoradiotherapy.

Abstract

Nasopharyngeal carcinoma (NPC) is a malignant tumor that targets the nasopharyngeal mucosal epithelium. Concurrent chemoradiotherapy (CCRT) is a pivotal treatment modality for NPC, yet it poses a risk for osteoradionecrosis (ORN), a complication that can impede further treatment. This study sought to explore the risk factors for ORN in NPC patients post-CCRT and to construct predictive models. We performed a retrospective analysis of clinical data from 417 NPC patients treated with CCRT at the Affiliated Hospital of Jiangnan University, with 204 patients from Longyan First Hospital as a validation cohort for the models. Our findings indicated that a high radiation dose, tooth extraction after radiotherapy, inadequate oral hygiene, smoking, anemia, and advanced T staging were associated with an elevated risk of ORN in NPC patients following CCRT. We formulated risk prediction models for ORN utilizing a nomogram, gradient boosting machine (GBM), and random forest (RF) algorithms. The area under the curve (AUC) was 0.813 (95% CI: 0.724-0.902) for the nomogram model in the validation cohort, 0.821 (95% CI: 0.732-0.910) for the GBM, and 0.735 (95% CI: 0.614-0.855) for the RF. Delong's test indicated no statistically significant differences in the AUC values among the three models. The nomogram has strong performance across both the training and validation cohorts, featuring a straightforward structure that is both intuitive and comprehensible. Taking into account the model's discriminative power, generalizability, and clinical practicability, the nomogram was proven to be highly applicable in the current study.