[Dosimetric factor analysis of the incidence of post-radiotherapy pneumonia after partial resection of laryngeal cancer and construction of a risk prediction nomogram model].

Abstract

Objective: To explore the influencing factors for pneumonia occurrence within three months in patients undergoing partial laryngectomy combined with radiotherapy, and to construct a nomogram prediction model for the risk of pneumonia occurrence. Methods: A total of 165 patients with laryngeal squamous cell carcinoma who underwent partial laryngectomy combined with postoperative radiotherapy in the Affiliated Cancer Hospital of Zhengzhou University from 2017 to 2023 were divided into three groups: a group without pneumonia, a low risk pneumonia group, and a medium to high risk pneumonia group according to the occurrence of pneumonia. The receiver operating characteristic (ROC) curve was used to determine the optimal cutoff values of the mean dose to the larynx (Larynx-Dmean) and the mean dose to the pharyngeal constrictor muscles (PCM-Dmean) for predicting the occurrence of pneumonia. Logistic regression analysis was used to screen out the influencing factors of pneumonia within three months after partial laryngectomy combined with radiotherapy, and a nomogram prediction model was constructed. Results: Among the 165 patients, 59 were in the group without pneumonia, 60 were in the group with a low risk of pneumonia, and 46 were in the group with a medium to high risk of pneumonia. The overall incidence of pneumonia was 64.2% (106/165). ROC curve analysis showed that the area under the curve (AUC) of the Larynx-Dmean for predicting the occurrence of pneumonia was 0.876, and the optimal cutoff value was 60.8 Gy. When the Larynx-Dmean was >60.8 Gy, the incidence of pneumonia was 96.7%. The AUC of the PCM-Dmean for predicting the occurrence of pneumonia was 0.747, and the optimal cutoff value was 54.6 Gy. When the PCM-Dmean was >54.6 Gy, the incidence of pneumonia was 74.7%. The AUC of the Larynx-Dmean for predicting the occurrence of medium to high risk pneumonia was 0.987, and the optimal cutoff value was 68.2 Gy. When the Larynx-Dmean was >68.2 Gy, the incidence of medium to high risk pneumonia was 100.0%. The AUC of the PCM-Dmean for predicting the occurrence of medium to high risk pneumonia was 0.850, and the optimal cutoff value was 58.7 Gy. When the PCM-Dmean was >58.7 Gy, the incidence of medium to high risk pneumonia was 85.2%. Multivariate logistic regression analysis showed that the incidence of pneumonia was higher in patients with a history of smoking (OR=1.54, 95% CI: 1.02-1.74), a higher smoking index (OR=2.43, 95% CI: 1.16-7.59), a higher Larynx-Dmean (OR=3.79, 95% CI: 1.25-6.49), and a higher PCM-Dmean (OR=2.44, 95% CI: 1.53-3.16). A nomogram prediction model for the risk of pneumonia and medium to high risk pneumonia within three months after partial laryngectomy combined with radiotherapy was successfully constructed, with C indices of 0.78 and 0.98, respectively, which had good predictive performance and clinical efficacy. Conclusions: When the Larynx-Dmean was >60.8 Gy and the PCM-Dmean was >54.6 Gy for postoperative radiotherapy after partial laryngectomy, the possibility of pneumonia occurrence in patients within three months increased significantly. Smoking history, smoking index, Larynx-Dmean, and PCM-Dmean were independent influencing factors for the occurrence of pneumonia. The nomogram prediction model constructed based on this can guide clinicians to take intervention measures to prevent the occurrence of pneumonia in patients with laryngeal cancer.