Diagnostic Accuracy of Frozen Section and Its Influence on Intraoperative Management of Indeterminate Epithelial Ovarian Tumors

Abstract

The objective of this study is to determine the diagnostic accuracy of frozen section in detecting epithelial ovarian tumor histological types and its effect on management. A retrospective review was done of all patients who had an intraoperative frozen section for an indeterminate ovarian tumor over a six-year period. The reference standard was final histology. The validity indices for frozen section in diagnosing benign, borderline, and malignant lesions were determined. One hundred thirty-five intraoperative frozen section–diagnosed epithelial ovarian tumors were reviewed. The mean age was 44.9 ± 14.2 years, the median parity was 2, and 57% (77/135) of patients were post-menopausal. The commonest histological subtype was mucinous 48.1% (65/135) on frozen section and 46.7% (63/135) on final histology. The overall concordance rate of frozen section to final histology was 81.5% (ƙ = 0.719, p = 0.0001). The accuracy, sensitivity, specificity, and positive predictive value of frozen section to diagnose benign lesions were 86.7%, 85.7%, 97.2%, and 79.2% respectively. In borderline tumors, the diagnostic test characteristics were 88.1%, 81.2%, 90.3%, and 72.2%. For malignant lesions, these values were 88.1%, 77.8%, 95.1%, and 91.3% respectively. The odds ratios for frozen section being correct were 40.9 (95% CI 14.8–113.5) for benign lesions, 40.3 (95% CI 13.4–121.3) for borderline tumors, and 67.4 (95% CI 20.5–222.0) for malignancy. Over-treatment or under-treatment occurred in 19.3% of patients. Intraoperative frozen section is useful in situations where the nature of the ovarian tumor is uncertain. However, borderline ovarian tumors are more likely to be over-diagnosed. About a fifth of patients received inappropriate treatments based on the frozen section report.