Development and Validation of a Clinical Prediction Model to Diagnose Sinonasal Inverted Papilloma Based on Computed Tomography Features and Clinical Characteristics

Abstract

Objectives: Sinonasal inverted papilloma (SNIP) is one of the most common benign tumors of the nasal cavity and sinuses and is at risk for recurrence and malignant transformation. It is crucial to precisely predict SNIP before surgery to determine the optimal surgical technique and prevent SNIP recurrence. This study aimed to evaluate the diagnostic value of computed tomography (CT) features and SNIP clinical characteristics and to develop and validate a clinically effective nomogram. Methods: Here, 267 patients with SNIP and 273 with unilateral chronic rhinosinusitis with/without nasal polyps were included. Patient’s demographic and clinical characteristics (i.e., gender, age, nasal symptoms, history of sinus surgery, smoking, and alcohol dependence) and CT features (i.e., lobulated/wavy edge, air sign, focal hyperostosis, diffuse hyperostosis, focal osseous erosion, and CT values) were recorded. Independent risk factors were screened using logistic regression analysis. A nomogram model was developed and validated. Results: Logistic regression analysis showed that age, facial pain/headache, history of sinus surgery, lobulated/wavy edge, air sign, focal hyperostosis, focal osseous erosion, and CT values were independent predictors of SNIP. A nomogram comprising these 8 independent risk factors was established. The area under the curve (AUC) for the training set was .960 (95% CI, .942–.978) and the AUC for the validation set was .951 (95% CI, .929–.971). Conclusion: The obtained results suggested that the nomogram based on age, facial pain/headache symptoms, history of sinus surgery, and CT characteristics had an excellent diagnostic value for SNIP.