Objective: A retrospective study was undertaken to evaluate the accuracy of frozen section diagnosis in gynecological surgery.
Methods: We compared the results of 792 consecutive gynecological frozen section diagnoses with their final diagnoses from January 1991 to June 1996. Slides for which the frozen section diagnosis was uncertain or incompatible with the final diagnosis were reviewed by an attending pathologist to determine the possible causes.
Results: A total of 299 ovarian, 390 lymph node, 56 uterine lesions, and 77 other tissue samples were obtained. The frozen section diagnosis was compatible with the final diagnosis in 97.5% of cases. The sensitivity for nonbenign lesions was 90.9%, and the specificity was 99.5%. There were no false positives or overestimated cases; 1.3% of cases were falsely negative, 0.4% underestimated the degree of malignancy, and 0.9% were uncertain. Possible causes for incompatible or uncertain frozen section diagnoses were analyzed. The accuracy of frozen section diagnoses for ovarian, lymph node, uterine, and other tissues was also evaluated. Frozen section was found to identify correctly 13 of 17 ovarian malignancies metastaic from other organs, 14 of 15 germ cell malignancies, and 3 of 4 dysgerminomas. The low sensitivity in ovarian borderline malignancy was due to the even lower sensitivity in its mucinous subgroup. The relationship between section numbers and accuracy of frozen section diagnosis in mucinous ovarian tumors was assessed.
Conclusions: Frozen section diagnosis in gynecology is sufficiently accurate for clinical use, with a low false negative rate and an even lower false positive rate. Most incompatible frozen section diagnoses occurred in ovarian lesions, especially in mucinous ovarian tumors. Performing multiple sections (at least one section for every 10 cm in diameter) is recommended in the frozen section diagnosis of mucinous ovarian tumors.