Accuracy of frozen section diagnosis of borderline ovarian tumors

Abstract

Objective: To determine the correlation between the diagnosis of borderline ovarian tumors (BOTs) by frozen section and permanent histology analyses.

Methods: Three hundred fifty-four pathology reports with diagnoses of BOTs by frozen section or permanent histology analysis at a single institution between 1995 and 2010 were evaluated with a review of the literature. Frozen section and permanent histology analyses were compared. Multivariate regression analysis was used to assess the influence of clinicopathological parameters on the likelihood of underdiagnosis.

Results: The overall accuracy, i.e., agreement between frozen section and permanent histology diagnoses, was observed in 228 of 354 (64.4%) cases, yielding a sensitivity of 72.6%, a positive predictive value of 85.1%, underdiagnosis in 108 cases (30.5%), and overdiagnosis in 18 cases (5.1%). Based on multivariate analysis, mucinous histology (OR, 1.48; P=0.022) was the only significant predictor for underdiagnosis by frozen section. A comprehensive search of the literature identified 46 studies investigating the accuracy of frozen section analysis of BOTs. The data of 7 of 46 studies that met the criteria for inclusion and the data of the current study were pooled. The overall accuracy was 67.1% (741/1104), yielding a sensitivity of 82.1%, a positive predictive value of 78.7%, underdiagnosis in 222 cases (20.1%), and overdiagnosis in 141 cases (12.8%).

Conclusions: Frozen section analysis of BOTs has low accuracy, sensitivity, and positive predictive value, and underdiagnosis and overdiagnosis are frequent. Therefore, surgical decision-making for BOTs based on frozen section diagnosis should be done carefully, especially in tumors with mucinous histology.