The evaluation of new therapies in prostate cancer requires unique end points for agents with diverse mechanisms of action. Because retinoic acid may have a confounding effect on prostate-specific antigen, we incorporated a pathological end point into the outcome assessment of two sequential clinical trials using all-trans-retinoic acid (ATRA) and the combination of 13-cis-retinoic acid and IFN-2a (cRA¿IFN). Pre- and posttherapy tumor biopsy specimens were studied for histological changes, apoptosis (terminal deoxynucleotidyl transferase-mediated nick end labeling assay), and proliferation index (Ki67). Prostate-specific membrane antigen (PSMA) expression was also evaluated using two different monoclonal antibodies to its intracellular domain (Cytogen 7E11 and Hybritech PM2). Fourteen patients with androgen-independent disease were treated with ATRA (50 mg/m2 p.o. every 8 h daily) and 16 androgen-independent and 4 androgen-dependent patients were treated with cRA¿IFN (10 mg/kg/day cRA plus 3, 6, or 9 million units daily IFN). Both therapies were well tolerated, with fatigue and cheilitis being the most common adverse events. Clinical activity, assessed by radiographs and serum prostate-specific antigen, was minimal, and the majority of patients progressed within 3 months. One patient with androgen-dependent disease had prolonged stabilization for >1 year. The majority of cases (95%) showed no gross histological changes and no difference in apoptotic or proliferative indices. Increased PSMA immunoreactivity was seen in seven of nine (78%) cases using PM2 antibody and in two of nine (22%) cases using the 7E11 antibody. Although antitumor effects were modest, the results suggest a role for retinoids in modulating the expression of PSMA on prostate cancer cells.