• prostate;
  • neoplasm;
  • radical prostatectomy;
  • staging;
  • PSA;
  • PSA density;
  • biochemical recurrence;
  • prognosis


To investigate the relationship of preoperative prostate-specific antigen (PSA) level and PSA density with several clinical and pathological variables, including biochemical recurrence after radical prostatectomy (RP), and to compare the preoperative PSA level and PSA density as prognostic factors in prostate cancer.


The study included 348 patients who had a RP at one institution, with whole-mount specimens of the prostate examined by one pathologist. Univariate and multivariate analyses were used to assess the relationship of the preoperative PSA level and PSA density with clinical and pathological variables, and by receiver operating characteristic (ROC) analysis to evaluate the relative usefulness of the two factors as predictors for biochemical recurrence.


The PSA level before RP was significantly correlated (Spearman's rank correlation) with patient age (P = 0.003), prostate weight (P < 0.001), cancer volume (P < 0.001) and Gleason score (P = 0.033), and with surgical margin status and pathological stage (both P < 0.001) in the RP specimen. In the multivariate analysis controlling for tumour stage, surgical margin status, and Gleason score, both PSA level and PSA density were significant predictors of PSA recurrence (P = 0.027 and 0.01, respectively). ROC analysis showed no statistical difference between the PSA level and PSA density in predicting PSA recurrence after RP (P = 0.40).


These results show a significant correlation of the preoperative PSA level with other established prognostic factors for prostate cancer. In the multivariate analysis, both PSA level and PSA density were independent predictors of PSA recurrence. Because the PSA level is as effective as PSA density in predicting PSA recurrence, the extra effort required to calculate PSA density may not be warranted. We recommend that the PSA level before RP be considered in stratifying patients into different prognostic groups, and in determining the optimum management.