Intramedullary spinal cord ependymoma and astrocytoma: Intraoperative frozen-section diagnosis, extent of resection, and outcomes

Abstract

OBJECTIVE The intraoperative differentiation of ependymomas from astrocytomas is important because neurosurgical strategies differ between these two tumor groups. Previous studies have reported that the diagnostic accuracy of intraoperative frozen sections of intracranial central nervous system (CNS) tumors is higher than 83%–97%, whereas that for spinal intramedullary tumors remains unknown. Herein, authors tested the hypothesis that intraoperative frozen-section diagnosis is the gold standard for a differential diagnosis of intramedullary spinal cord tumors. METHODS The clinical characteristics, intraoperative histological diagnosis from frozen sections, extent of tumor resection, progression-free survival (PFS), and overall survival (OS) of 49 cases of intramedullary spinal cord ependymomas (n = 32) and astrocytomas (n = 17) were retrospectively evaluated. RESULTS The frozen-section diagnosis and final diagnosis with permanent sections agreed in 23 (72%) of 32 cases of ependymoma. Of the 9 cases of ependymoma in which the frozen-section diagnosis disagreed with the final diagnosis, 4 were incorrectly diagnosed as astrocytoma and the other 5 cases had a nonspecific diagnosis, such as glioma. Nonetheless, gross-total resection was achieved in 6 of these 9 cases given the presence of a dissection plane. The frozen-section diagnosis and final diagnosis agreed in 12 (71%) of 17 cases of astrocytoma. Of the 5 cases of astrocytoma in which the frozen-section diagnosis disagreed with the final diagnosis, 1 was incorrectly diagnosed as ependymoma and the other 4 had a nonspecific diagnosis. Gross-total resection was achieved in only 1 of these 5 cases. A relationship between the size of tumor specimens and the diagnostic accuracy of frozen sections was not observed. Ependymal rosettes and perivascular pseudorosettes were observed in 30% and 57% of ependymomas, respectively, but were absent in astrocytomas. Progression-free survival and OS were both significantly longer in cases of ependymoma than in cases of astrocytoma (p < 0.001). Gross-total resection was achieved in 69% of ependymomas and was associated with longer PFS (p = 0.041). In the astrocytoma group, gross-total resection was achieved in only 12% and there was no relationship between extent of resection and OS. Tumor grades tended to correlate with OS in astrocytomas (p = 0.079). CONCLUSIONS The diagnostic accuracy of intraoperative frozen sections was lower for intramedullary spinal cord ependymomas and astrocytomas in the present study than that for intracranial CNS tumors reported on in the literature. Surgical strategies need to be selected based on multiple factors, such as clinical characteristics, preoperative imaging, frozen-section diagnosis, and intraoperative findings of the tumor plane.