Is intraoperative frozen section analysis an efficient way to reduce positive surgical margins?

OBJECTIVES: To assess the accuracy and efficiency of frozen section analysis to detect positive surgical margins (+SMs) during radical prostatectomy.
METHODS: In a consecutive series of 760 patients treated with radical prostatectomy from 1998 to 2002, areas suspicious for +SMs on the surface of the removed prostate were examined by frozen section analysis. In a subset of 520 patients, the surgeon's level of suspicion for +SMs was scored and recorded during radical prostatectomy.
RESULTS: Overall, 259 patients underwent frozen section examination. Of these, 55 patients (21%) had +SMs on permanent section examination compared with 50 (10%) of 501 patients with no frozen section analysis (P <0.005). Cancer was present in 23 (8.9%) frozen section specimens, all of which were confirmed on permanent section analysis. Frozen section examination missed 32 +SMs in 236 patients with negative frozen section results. The sensitivity, specificity, and positive and negative predictive value of frozen section analysis to identify +SMs was 42%, 100%, 100%, and 86%, respectively. However, the sensitivity of frozen section analysis was much lower (23 of 105, 22%) when analyzed for the entire population, including those who did not have frozen section analysis. Among the subset of 520 patients with the level of suspicion recorded, 79 had a +SM on permanent section examination. However, 51 (64%) of these were in patients with no suspicious area in the prostatectomy specimen.
CONCLUSIONS: Although the positive predictive value of frozen section analysis for +SMs is high, the sensitivity is too low to expect that a policy of routine frozen section analysis of suspicious areas will reduce the rate of +SMs substantially.