Accuracy of frozen section in distinguishing primary ovarian neoplasia from tumors metastatic to the ovary.

Frozen section is widely used in the intra-operative assessment of patients with ovarian tumors. The diagnosis of malignancy is usually straightforward but in some cases it may be difficult to distinguish whether tumors are of ovarian origin or represent matastases from other sites. Recently, Seidman and colleagues presented a simple algorithm based on tumor size and unilateral versus bilateral involvement to aid in intra-operative assessment of ovarian mucinous neoplasms. In this study we have reviewed the accuracy of frozen section in distinguishing primary ovarian malignancies from tumors metastatic to the ovaries encountered in two hospitals over a 5-year period. The algorithm was also applied to our cases retrospectively irrespective of histological type. Nine hundred fourteen ovarian frozen sections were performed in the study period including 266 cases with a final diagnosis of malignancy. Thirty-seven malignancies (13.9%) were of metastatic origin (exclusing one lymphoma), 21 of which (58.8%) were correctly identified on frozen section. In 5 additional cases metastatic origin was included in the differential diagnosis while a primary ovarian tumor was favored un 11 cases (29.7%). Application of the algorithm to the metastatic tumors led to correct classification in 26/33 (78.8%) assessable cases. Conversely, 195/228 primary ovarian malignancies were correctly identified intra-operatively but the possibility of extra-ovarian malignancy was considered or not excluded in 33 cases (14.5%). Application of the algorithm to the latter problematic primary ovarian tumors overall was not helpful in distinguishing primary or metastatic origin. However if only low-grade primary adenocarcinomas were considered then 10/12 assessable cases were correctly assigned. In conclusion frozen section is only moderately successful in distinguishing primary ovarian malignancies fron tumors metastatic to the ovaries. The simple algorithm proposed by Seidman and colleagues for assessment of ovarian mucinous tumors is helpful and can be applied to low-grade adenocarcinomas of other histological types.