Percentage of positive biopsy cores can improve the ability to predict lymph node invasion in patients undergoing radical prostatectomy and extended pelvic lymph node dissection

Objective: We hypothesized that the information stemming from biopsy cores can enhance the ability to predict the rate of lymph node invasion (LNI) at radical retropubic prostatectomy (RRP) in men subjected to extended pelvic lymphadenectomy (ePLND). Materials and methods: A cohort of 278 consecutive patients (mean age: 66.2 yr) underwent a RRP and an ePLND, in which 10 or more nodes were removed and examined. The median PSA was 7.5 ng/ml. Clinical stage was mostly T1c (59.4%) and T2 (37.8%). Biopsy Gleason sum was 2-5 in 26.6%,6 in 39.2%, 7 in 27%, and 8-10 in 7.2%. The number of positive cores was 1-19 (median: 4), whilst percentage of positive cores was 7.1-100% (median: 37.5%). Logistic regression models tested the association between the above predictors and LNI. Testing of PSA was coded as either a continuous variable (CV) or a cubic spline (CS). Individual variables and combined accuracy were tested in regression-based nomograms, which were subjected to 10,000 bootstrap resamples to reduce overfit bias. Results: Mean number of lymph nodes examined was 17.5 (range: 10-38); 29 patients (10.4%) had LNI. Percentage of positive cores (78.5%) and biopsy Gleason sum (78.4%) were the most informative predictors of LNI. A nomogram based on clinical stage, PSA (CV), and biopsy Gleason sum was 79.7% accurate versus 83% (3.3% gain, p < 0.001) when percentage of positive cores was added. A 2.7% gain (83.7% vs. 81%; p < 0.001) was recorded after the addition of the percentage of positive cores when PSA was coded as a CS. Conclusions: Percentage of positive biopsy cores should be considered in prediction of LNI at ePLND, because it significantly improves the combined accuracy of established clinical predictors such as PSA, clinical stage, and biopsy Gleason sum. (C) 2007 European Association of Urology. Published by Elsevier B.V. All rights reserved.