Diagnostic accuracy of frozen section of central nervous system lesions: a 10-year study

Definitive diagnosis of the central nervous system (CNS) lesions is unknown prior to histopathological examination. To determine the method and the endpoint for surgery, intraoperative evaluation of the lesion helps the surgeon. In this study, the diagnostic accuracy and pitfalls of using frozen section (FS) of CNS lesions is determined.

In this retrospective study, we analyzed the results of FS and permanent diagnoses of all CNS lesions by reviewing reports from 3 general hospitals between March 2001 and March 2011.

273 cases were reviewed and patients with an age range from 3 to 77 years of age were considered. 166 (60.4%) had complete concordance between FS and permanent section diagnosis, 83 (30.2%) had partial concordance, and 24 cases (9.5%) were discordant. Considering the concordant and partially concordant cases, the accuracy rate was 99.5%, sensitivity was 91.4%, specificity was 99.7%, and positive and negative predictive values were 88.4% and 99.8%, respectively.

Our results show high sensitivity and specificity of FS diagnosis in the evaluation of CNS lesions. A Kappa agreement score of 0.88 shows high concordance for FS results with permanent section. Pathologist's misinterpretation, small biopsy samples (not representative of the entire tumor), suboptimal slides, and inadequate information about tumor location and radiologic findings appear to be the major causes for these discrepancies indicated from our study.