Accuracy of frozen section in the diagnosis of malignant ovarian tumor.

Abstract

To evaluate the diagnostic accuracy of frozen section for histopathologic diagnosis of ovarian tumors. A total of 147 surgically removed ovarian tumors were studied. Each ovarian tumor sample was evaluated for histopathologic diagnosis using both frozen and paraffin sections. Interpretation was separate and blinded between each technique. Accuracy, diagnostic values and their 95% confidence intervals (CI) were estimated by comparing the results from both techniques, using paraffin section as a gold standard. Overall accuracy of frozen section was 89.8% (95% CI 83.4-94.0). Sensitivity was 90.4% (95% CI 78.2-96.4) for malignant, 33.3% (95% CI 6.0-75.9) for borderline, and 93.3% (95% CI 85.4-97.2) for benign tumors. The predictive value was 100% (95% CI 90.6-100) for malignant, 20% (95% CI 3.5-55.8) for borderline, and 92.2% (95% CI 84.1-96.5) for benign tumors. Most false negatives occurred in mucinous and borderline tumors. No benign tumor was misdiagnosed as malignant by frozen section. Accuracy and negative predictive value were significantly lower in epithelial rather than germ and other cell types. Excellent agreement with regard to histologic cell type was observed (Kappa 0.81). Frozen section appears to be an accurate technique for the histopathologic diagnosis of ovarian tumors. Some limitations were observed among borderline and mucinous tumors; this emphasizes the great value of frozen section in the diagnosis of ovarian tumors.