Radiological Findings of Medial Orbital Wall Bony and Periorbital Dehiscence in Sinonasal Malignancies as a Predictor of Final Histopathologic Orbital Invasion

Background: Malignant tumors of the nasal cavity and paranasal sinuses have the potential to invade into surrounding structures including the orbit and skull base. The orbital periosteum has long been considered a barrier to the spread of such tumors into the orbital cavity and can be regarded as the decisive layer when determining extent of tumor resection. However, interpretation of orbital invasion can become extremely challenging as the periosteum can be difficult to visualize separately from the orbital bony wall on imaging.

Purpose: This study aims to determine the impact of computed tomography findings and operative assessment of orbital bony and periosteal tumor involvement as an accurate predictor of final microscopic invasion.

Methods: A single-center, retrospective review of 188 patients undergoing surgical resection of sinonasal malignancy between 2008 and 2019 was performed. Patients were included if they had a sinonasal malignancy involving the orbit with the appropriate radiographic, operative, and final pathologic reports available.

Results: A total of 82 cases met inclusion criteria. All patients underwent either endoscopic or open resection for invasive sinonasal malignancy. Treatment was by orbital exenteration in 27 patients (33%) and orbital preservation in 55 patients (67%).

A total of 26 patients (31.7%) were found to have medial orbital bony wall dehiscence on preoperative CT imaging. Of these patients, 14 were noted to have bone involvement on operative assessment and 7 patients had positive orbital wall bone on final pathology. Forty-nine (59.7%) patients had periorbital involvement on imaging, 19 of who were noted on operative assessment and 13 with final positive pathology. Periorbital involvement was defined as gross tumor extension into the orbit, suggesting breach.

Erosion of the orbital bony wall and periorbita on preoperative CT was found to be 70% [95% CI 35.4%, 91.9%] and 28% [95% CI 12.9%, 50%] sensitive and 69.8% [95% CI 56.8%, 80.4%] and 62% [95% CI 46.4%, 75.1%] specific respectively for involvement on final pathology. The positive predictive values (PPV) were 26.92 and 33.3% and the negative predictive values (NPV) were 93.62 and 63.4%, respectively.

Operative assessment of bony erosion and periorbital involvement was found to be 60% [95% CI 27.4%, 86.3%] and 45.8% [95% CI 26.2%, 66.8%] sensitive and 68.7% [95% CI 56.0%, 79.1%] and 59.6% [95% CI 45.1%, 72.7%] specific, respectively, for invasion on final pathology. The PPVs were 22 and 34.4% and NPVs were 92 and 70.45%, respectively.

Overall, it is more likely that a patient with pathologic evidence of bony involvement had bony erosion on operative assessment (OR: 3.28, p < 0.05).

Conclusion: Radiologic findings and intraoperative assessment for bony and periorbital erosion are valuable tools when assessing orbital margins, but neither are perfect. This re-enforces the need for intraoperative pathological consultation when attempting to preserve critical structures. Additionally, surgeons should therefore be aware of the potential need for aggressive tumor resection within the orbit in cases with low preoperative probability of invasion.