To evaluate the prognostic potential of prostate-specific antigen (PSA) expression in tumor specimens in a large cohort of prostate cancers treated by prostatectomy. Although serum PSA measurement has been established as a diagnostic and prognostic tool in prostate cancer, no larger studies have been done of the prognostic potential of this parameter.

We used a tissue microarray containing samples of 3261 prostatectomy specimens. PSA expression was scored after immunohistochemical staining on a scale from 0 (absent) to 3 (strong) by an investigator unaware of all other variables. The results were correlated with the pre- and postoperative clinical and pathologic parameters and follow-up data.

Of 2556 eligible tumors, PSA expression was strong in 48.0%, moderate in 36.7%, weak in 12.2%, and absent in 3.1%. The loss of PSA expression correlated significantly with a greater Gleason score, the presence of extraprostatic extension, and a peripheral zone prostate cancer location. It was also significantly associated with PSA recurrence after prostatectomy on univariate analysis but not on multivariate analysis containing the pathologic parameters of the prostatectomy specimens. In the subsets of patients with a preoperative PSA value <6 ng/mL or biopsy Gleason score of 3 + 4, the loss of PSA expression in tissue microarray spots was significantly associated with nonorgan-confined disease.

The results of our study have shown that the loss of PSA expression in tissue samples of prostate cancer is associated with adverse pathologic features and clinical outcome but is not an independent prognostic factor for PSA recurrence after prostatectomy. However, in the biopsy scenario and in subgroups of patients, it might be a useful parameter for predicting extraprostatic tumor extension.