Purpose: An algorithm including the results of systematic sextant biopsies was statistically developed and evaluated to predict the probability of pelvic lymph node metastases in patients with clinically localized carcinoma of the prostate.
Materials and methods: Clinical stage, serum prostate specific antigen concentration, Gleason score, number of positive biopsies, number of biopsies containing any Gleason grade 4 or 5 cancer and number of biopsies predominated by Gleason grade 4 or 5 cancer were recorded in 345 patients undergoing pelvic lymph node dissection and correlated with the incidence of lymph node metastases. Multivariate logistic regression, and classification and regression trees analyses were performed.
Results: In univariate analysis all variables had a statistically significant influence on lymph node status. Logistic regression showed that the amount and distribution of undifferentiated Gleason grade 4 and 5 cancer in the biopsies were the best predictors of lymphatic spread followed by serum prostate specific antigen. Classification and regression trees analysis classified 79.9% of patients who had 3 or fewer biopsies with Gleason grade 4 or 5 cancer and no biopsies predominated by undifferentiated cancer as a low risk group. In this group positive lymph nodes occurred in only 2.2% (95% confidence interval 0.8 to 4.7%).
Conclusions: Including the results of systematic sextant biopsies substantially enhances the predictive accuracy of algorithms that define the probability of lymph node metastases in prostatic cancer. Patients thus defined as having no lymphatic spread could potentially be spared pelvic lymph node dissection before definitive local treatment.