2026 Poster Presentations
P308: CORRELATION BETWEEN INTRAOPERATIVE FROZEN AND FINAL DURAL MARGINS IN SINONASAL AND VENTRAL SKULL BASE MALIGNANCIES
Michael Xie, MD; Diana Bell, MD; Garret W CHoby, MD; Georgios Zenononos, MD; Carl H Snyderman, MD, MBA; Paul A Gardner, MD; Eric W Wang, MD; University of Pittsburgh Medical Center
Background: Sinonasal and anterior skull base malignancies can abut or invade adjacent dura. Dural involvement and positive dural margins portend negative oncologic outcomes. Even malignancies that encroach the skull base without radiographic involvement, occult dural invasion remains high. The extent of dural resection in gross or suspected dural involvement must be balanced against the increased risk to critical neurovascular structures, reconstructive complexity, and cerebrospinal fluid leaks. Intraoperative frozen margins are used to guide intraoperative decision making. Discordances between frozen and final pathology occur and have been demonstrated in other settings. To date, no data characterizes the diagnostic characteristics of intraoperative frozen assessment for dural margins in sinonasal and anterior skull base malignancies.
Objective: To characterize the diagnostic characteristics of dural frozen margins compared to final pathology among adult patients with sinonasal and anterior skull base malignancies.
Methods: This was a single institution retrospective chart review. All sinonasal and anterior skull base malignancies treated surgically at a single center from 2010-2025 were reviewed. Eligibility criteria included adult patients (≥ 18yo), pathologic diagnosis of sinonasal or anterior skull base malignancies, treated via endoscopic endonasal approach (EEA) or combined (open/endoscopic) approach involving dural resection, and complete frozen and final pathology data. Patient, tumor, treatment, and pathology data were extracted. Individual dural frozen and final specimens (per specimen analysis) as well as tumor dural frozen and final margin statuses (per tumor analysis) were compared to determine diagnostic test characteristics.
Results: Out of 316 sinonasal and anterior skull base malignancies, 69 patients met the study eligibility criteria with a total of 404 frozen dural margins. Histopathologies included 32 olfactory neuroblastoma (ONB), 11 squamous cell carcinomas (SCC), 7 adenocarcinomas, 6 sinonasal undifferentiated carcinomas, 5 sinonasal neuroendocrine carcinomas (SNEC), and 8 other malignancies. 59 patients were treated via EEA and 10 patients via combined approach. There were 2 false positives, 10 false negatives, 34 indeterminate results, and 5 frozens without final pathology due to tissue exhaustion during processing. Per specimen analysis showed sensitivity, specificity, positive predictive value, and negative predictive value of 0.78, 0.99, 0.95, 0.97, respectively. Per tumor analysis showed a sensitivity, specificity, positive predictive value, and negative predictive value of 0.29, 0.93, 0.57, 0.81, respectively. Per specimen sensitivity analysis interpreting indeterminate results as positive, showed sensitivity, specificity, positive predictive value, and negative predictive value of 0.79, 0.93, 0.64, 0.96, respectively. Highest rates of frozen-final discordance and indeterminate specimens were observed in SCC and neuroectodermal tumors (ONB and SNEC). False positive frozens were only observed in ONB.
Conclusions: Frozen dural margins have high specificity in the setting of individual specimens and overall tumor margins. Negative and indeterminate frozen margins are less accurate and should be interpreted as a supplement to other preoperative and intraoperative factors during surgical decision making. Surgeons and pathologists should be cognizant of the limitations of dural frozens due to inaccuracy, indeterminate results, and issues with tissue processing.