Frozen section diagnosis of metastatic prostate adenocarcinoma in pelvic lymphadenectomy compared with nomogram prediction of metastasis

Objectives. To compare the sensitivity and negative predictive values of frozen section analysis of pelvic lymphadenectomy in patients undergoing radical retropubic prostatectomy for prostate adenocarcinoma with the predictive power of published nomograms for metastasis to lymph nodes. Methods. A retrospective review was performed on all patients who underwent bilateral pelvic lymphadenectomy and radical retropubic prostatectomy for prostate adenocarcinoma between 1991 and early 1997. The sensitivity and negative predictive values were computed comparing frozen section analysis, and patients were grouped by risk stratification. Comparison was made using the McNemar text. Results. The sensitivity for detecting lymph node metastasis on frozen section analysis for all risk groups was 33% (9 of 27). The sensitivity for identifying patients at high risk of having nodal metastasis by published nomograms alone was 67% (18 of 27) (P = 0.04). The overall negative predictive value for frozen section analysis was 96.5% (503 of 521). The negative predictive value for uninvolved lymph nodes, using low and intermediate-risk groups stratified by published nomograms, was 97.9% (436 of 445). Conclusions. Frozen section analysis of pelvic lymph nodes to detect metastatic prostate adenocarcinoma is less sensitive in determining which patients will have lymph nodes involved by metastatic adenocarcinoma than using risk stratification by published nomograms. The negative predictive value of frozen section analysis in all risk groups was very high, up to 97.9%. (C) 2002, Elsevier Science Inc.