Accuracy of frozen section diagnosis in soft tissue tumors

This study was undertaken to analyze the accuracy of frozen section (FS) diagnosis of 118 soft tissue tumors with respect to the reasons for which the intraoperative consultation was indicated. Fifty-seven frozen sections were performed for the diagnosis of an unknown pathologic process. Complete agreement was established in 40.3% and the correct pathologic process in 43.9%, the diagnosis was deferred in 14%, and the remaining 1.8% were diagnosed incorrectly. Examination for determination of the adequacy of resection margin (22 cases), lymph node or skip metastases (23 cases), residual or recurrent tumor after previous surgery (29 cases), viable tumor tissue after previous locoregional or systemic therapy (ten cases), and identification of the specimen (five cases) proved to be 95.5%, 95.7%, 96.6%, 90%, and 100% accurate. Considering the whole series, an erroneous answer to a question posed by a surgeon was given in four cases (two false positive and two false negative), of which two cases were a sampling error made by pathologist. Intraoperative consultation by FS in soft tissue tumors is (a) reliable for general rather than exact diagnosis in defining the previously unknown pathologic process and (b) mandatory in evaluating resection margins and any discrepancies between preoperative cytologic and intraoperative gross impression.