Abstract
BACKGROUND: Immune-related pneumonitis is a rare and potentially fatal adverse event associated with sintilimab. We aimed to develop and validate a nomogram for predicting the risk of immune-related pneumonitis in patients treated with sintilimab. METHODS: The least absolute shrinkage and selection operator (LASSO) regression was used to determine risk factors. Multivariable logistic regression was used to establish a prediction model. Its clinical validity was evaluated using calibration, discrimination, decision, and clinical impact curves. Internal validation was performed against the validation set and complete dataset. RESULTS: The study included 632 patients; 59 were diagnosed with immune-related pneumonitis. LASSO regression analysis identified that the risk factors for immune-related pneumonitis were pulmonary metastases (odds ratio [OR], 4.015; 95% confidence interval [CI]: 1.725-9.340) and metastases at >3 sites (OR, 2.687; 95% CI: 1.151-6.269). The use of combined antibiotics (OR, 0.247; 95% CI: 0.083-0.738) and proton pump inhibitors (OR, 0.420; 95% CI: 0.211-0.837) were protective factors. The decision and clinical impact curves showed that the nomogram had clinical value for patients treated with sintilimab. CONCLUSIONS: We have developed and validated a practical nomogram model of sintilimab-associated immune-related pneumonitis, which provides clinical value for determining the risk of immune-related pneumonitis and facilitating the safe administration of sintilimab therapy.