The prostatic-specific antigen (PSA) is the tumor marker most widely relied upon for the monitoring of patients with prostate cancer. Recently, declines in the serum concentrations of PSA have been advocated as a surrogate marker of tumor response in clinical trials of investigational antitumor agents. We examined the hypothesis that this postulate may not apply to the evaluation of drugs such as phenylacetate, a differentiating agent endowed with mechanisms of action different from those of classic cytotoxic chemotherapy. Using human prostatic carcinoma LNCaP cells as a model, we show that phenylacetate induces PSA production despite inhibition of tumor cell proliferation. Incubation of LNCaP cultures with cytostatic doses of phenylacetate (3-10 mM) resulted in a three- to fourfold increase in PSA secretion per cell. This appears to result from upregulation of PSA gene expression, as indicated by elevated PSA mRNA steady-state levels in treated cells. The increase in PSA production per cell was confirmed in rats bearing subcutaneous LNCaP tumor implants that were treated systemically with phenylacetate. Further comparative studies indicate that upregulation of PSA is common to various differentiation inducers, including all-trans-retinoic acid, 1,25-dihydroxyvitamin D3, and butyrate but is not induced by other antitumor agents of clinical interest such as suramin. We conclude that declines in PSA may be treatment specific and that the exclusive use of this criterion as a marker of disease response may mislead the proper evaluation of differentiating agents in prostate cancer patients.