Computed Tomography to Determine Sinonasal Inverted Papilloma Origin, Skull Base Involvement, and Stage

Objective: Sinonasal inverted papillomas (IP) are benign neoplasms that exhibit a tendency for local destruction, recurrence, and malignant transformation. Preoperative radiologic assessment with computed tomography (CT) is becoming increasingly more common to evaluate the origin and extent of tumor for surgical planning. This study aims to investigate the diagnostic ability of CT to determine the tumor origin, involvement of the skull base, and stage.

Design: Retrospective cross-sectional study.

Setting: Quaternary academic medical center

Participants: Fifty-two patients with preoperative CT imaging who have undergone extirpative surgery for histologically confirmed sinonasal IP.

Main Outcome Measures: Axial and coronal CT images in 1 mm slices were retrospectively reviewed by a neuroradiologist. The likely sites of tumor origin and skull base involvement based on opacification and surrounding osteitic changes were then determined based on imaging ([Fig. 1]). Clinical Krouse and Cannady stages were also determined from the CT images. These radiologic findings were then compared with the actual intraoperative and pathologic findings.

Results: In total, 76.9% (n = 40) of these sinonasal IP lesions on CT were concordant with the actual sites of tumor origin ([Fig. 2]). Determination of tumor origin via CT compared with intraoperative assessment revealed substantial agreement (κ= 0.70). In addition, CT had 60.0% sensitivity (95% CI: 26.2–87.8%), 90.5% specificity (95% CI: 77.4–97.3%), 60.0% positive predictive value (95% CI: 34.2–81.3%), 90.5% negative predictive value (81.6–95.3%), and 84.6% accuracy (95% CI: 71.9–93.1%) for diagnosing skull base involvement. Using the Krouse and Cannady staging system, clinical staging was concordant with pathologic staging in 73.1% (n = 38) and 76.9% (n = 40) of patients, respectively. There was moderate agreement between the clinical and pathologic Krouse (weighted κ = 0.59) and Cannady staging systems (weighted κ = 0.58).

Conclusion: Preoperative CT demonstrates substantial ability to determine the actual site of origin of sinonasal IP. While not sensitive, CT is highly specific and accurate for determining skull base involvement by tumor. CT also had a 19% probability of over-predicting tumor stage. Overall, CT appears to be a useful adjunct for surgical planning for sinonasal IP.

Fig. 1 Coronal CT image revealing focal hyperostosis of the left lateral lamella and fovea ethmoidalis, indicating skull base involvement.

Fig. 2 Distribution of sites of tumor origin based on computed tomography (CT) and intraoperative findings.