Purpose: We developed a preoperative nomogram for prediction of lymph node metastases in patients with clinically localized prostate cancer.
Materials and methods: The study was a retrospective, nonrandomized analysis of 7,014 patients treated with radical prostatectomy at 6 institutions between 1985 and 2000. Exclusion criteria consisted of preoperative androgen ablation therapy, salvage radical prostatectomy and pretreatment prostate specific antigen (PSA) greater than 50 ng/ml. Preoperative predictors of lymph node metastases consisted of pretreatment PSA, clinical stage (1992 TNM) and biopsy Gleason sum. These predictors were used in logistic regression analysis based nomograms to predict the probability of lymph node metastases.
Results: Overall 5,510 patients with complete clinical and pathological information were included in the study. Lymph nodes metastases were present in 206 patients (3.7%). Pretreatment PSA, biopsy Gleason sum, clinical stage and institution represented predictors of lymph node status (p <0.001). Bootstrap corrected predictive accuracy of the 3-variable nomogram (clinical stage, Gleason sum and PSA) was 0.76. Inclusion of a fourth variable, which accounts for institutional differences in lymph node metastases, yielded an area under the receiver operating characteristics curve of 0.78. The negative predictive value of our nomograms was 0.99 when they predicted 3% or less chance of positive lymph nodes.
Conclusions: Using clinical information, we produced 2 calibrated and validated nomograms, which accurately predict pathologically negative lymph nodes in men with localized prostate cancer who are candidates for radical prostatectomy.