We have hypothesized that some steroid derivatives bind to the androgen receptor (AR) with very low androgenic activity and therefore potentially function as better AR antagonists than clinically used antiandrogens, such as flutamide. Indeed, we previously found such a compound, 3b-acetoxyandrosta-1,5-diene-17-one ethylene ketal (ADEK), with some estrogenic activity. Here we report the identification of 2 additional steroid derivatives, 3b-hydroxyandrosta-5,16-diene (HAD) and androsta-1,4-diene-3,17-dione-17-ethylene ketal (OAK), as new potent antiandrogens. Like ADEK, HAD and OAK could interrupt androgen binding to the AR and suppress both dihydrotestosterone-and androstenediol-induced transactivations of wild-type and mutant ARs in prostate cancer cells. These 2 compounds also inhibited prostate-specific antigen expression in LNCaP as well as growth of different AR-positive prostate cancer cell lines stimulated by androgen. Significantly, HAD and OAK had only marginal agonist effects, as compared to hydroxyflutamide. More importantly, in contrast to ADEK, OAK was shown to possess marginal estrogenic activity. These results strengthen our hypothesis and suggest that selective steroid derivatives could be potent antiandrogenic drugs with less unfavorable effects for the treatment of prostate cancer. ' 2005 Wiley-Liss, Inc.