Accuracy of intraoperative pathological diagnosis using frozen sections of spinal cord lesions

Frozen sections are used to provide gross and rapid microscopic pathological information for guidance on intraoperative management and therapeutic decision-making. Many studies have shown the accuracy of frozen section diagnosis for intracranial lesions, but there are no studies focusing on spinal cord lesions. The purpose of this study is to evaluate the accuracy of intraoperative rapid diagnosis using frozen sections and to investigate limitations of this approach for spinal cord lesions.

We performed intraoperative pathological diagnosis using frozen sections in 67 cases in which it was difficult to determine malignancy in preoperative images. The lesions were extradural (ED) in 13 cases, intradural extramedullary (IDEM) in 27 cases, and intramedullary (IM) in 27 cases. Metastatic tumors were excluded. The accuracy of intraoperative pathological diagnosis and the patterns of incorrect diagnosis were examined.

Comparison of the intraoperative and final diagnoses gave an overall diagnostic sensitivity of 86.6% (58/67), with 100% (13/13) for ED lesions, 96% (26/27) for IDEM lesions, and 70% (19/27) for IM lesions. The diagnostic accuracy for IM lesions was significantly lower than those for ED and IDEM lesions (p < 0.05). Cases with small specimen sizes were frequently incorrectly diagnosed and inflammatory processes were common incorrect diagnoses using frozen specimens.

Among all spinal cord lesions, low diagnostic accuracy in intraoperative diagnosis using frozen sections is most likely for intramedullary lesions. The results of intraoperative rapid diagnosis should be interpreted with understanding of the limitations of this procedure.