Computed Tomography as a Predictor of Sinonasal Inverted Papilloma Origin, Skull Base Involvement, and Stage

• Published in: Journal of neurological surgery. Part B, Skull base Skull base, 2021-07, Vol.82 (S 03), p.e335-e341
• Main Authors: Lee, Jake J., Orlowski, Hilary L. P., Schneider, John S., Roland, Lauren T., Eldaya, Rami, Jiramongkolchai, Pawina, Kallogjeri, Dorina, Chernock, Rebecca D., Klatt-Cromwell, Cristine N.
• Format: Article
• Language: English
• Subjects: Original; Original Article
• ISSN: 2193-6331; 2193-634X
• DOI: 10.1055/s-0040-1701677

Abstract Objective  To investigate the diagnostic performance of computed tomography (CT) to determine the origin, skull base involvement, and stage of sinonasal inverted papilloma (IP). Design  This is a retrospective cohort study. Setting  This is set at a tertiary care medical center. Participants  Patients with preoperative CT imaging who underwent extirpative surgery for histologically confirmed sinonasal IP between January 2005 and October 2019. Main Outcome Measures  The likely sites of tumor origin, skull base involvement, and radiographic tumor stage were determined by two board-certified neuroradiologists after re-reviewing preoperative CT imaging. These radiologic findings were then compared with intraoperative and pathologic findings. Results  Of 86 patients, 74% (64/86) had IP lesions with correctly classified sites of origin on CT. CT was not sensitive for diagnosing ethmoid sinus origin (48%, 52%), frontal sinus origin (80%, 40%), and skull base origin (17%, 17%). CT was not sensitive (62%, 57%) but specific (86%, 98%) for identifying any skull base involvement. There was substantial-to-near perfect agreement between radiographic and pathologic Cannady stages (weighted κ = 0.61 for rater 1; weighted κ = 0.81 for rater 2). Interrater agreement was substantial for identifying tumor origin (κ = 0.75) and stage (weighted κ = 0.62) and moderate for identifying skull base involvement (κ = 0.43). Conclusion  Interrater agreement on CT findings was substantial except on skull base involvement. CT correctly predicted site of tumor origin in up to 74% of subjects. CT was not sensitive for diagnosing skull base involvement but had substantial-to-near perfect agreement with pathologic tumor staging. CT is a useful but albeit limited adjunct for tumor localization and surgical planning for sinonasal IP.