Prediction of malignant sinonasal inverted papilloma transformation by preoperative computed tomography and magnetic resonance imaging

Abstract

Background: Accurate preoperative prediction of the malignant transformation of sinonasal inverted papilloma (SNIP) is essential for radical resection of tumours and prevention of recurrence. We here explored the predictive value of preoperative computed tomography (CT) and magnetic resonance imaging (MRI). Methodology: The study was performed on 268 patients with SNIP with (n = 78) or without (n = 190) coexistent malignant transformation. We used univariate and multivariate logistic regression analysis method to screen for independent risk factors, and established a nomogram model. Finally, using receiver operating characteristic curves, we assessed the diagnostic value of the independent risk factors for malignant transformation of SNIP. Results: We identified bone erosion on CT, change in convoluted cerebriform pattern (CCP) on MRI, and washout-type time-intensity curve (TIC) of dynamic contrast-enhanced (DCE)-MRI were independent predictors of malignant transformation of SNIP. The scores of these three independent risk factors from the nomogram model were 10, 7 and 8, respectively. The area under the receiver operating characteristic curve for predicting SNIP malignancy was 0.954 for the nomogram model, 0.826 for bone erosion, 0.776 for washout-type TIC, and 0.810 for CCP mutation. Conclusions: Of the independent risk factors and related combination identified, the nomogram model based on bone destruction on CT, CCP mutation on MRI, and washout-type TIC of DCE-MRI had the best predictive value for malignant transformation of SNIP.