Hormones and Prostate Cancer: Where Do We Go From Here?

A striking phenomenon in prostate cancer epidemiology is that the prevalence of latent prostate tumors in most populations is similar (7), but the rates for clinical prostate cancer in U.S. men are 30-50 times higher than those for Asian men (7). Reasons for these disparities are unclear, although previous studies (7-5) have suggested racial differences in dietary fat intake, circulating levels of testosterone, intraprostatic activity of 5-alpha reductase (the enzyme that converts testosterone to the more potent dihydrotestosterone [DHT]), and polymorphisms of the androgen receptor (AR) gene (the intracellular receptor that binds DHT to form the DHT-AR complex that then binds to DNA to activate cell proliferation). Other than age, race, and a family history of prostate cancer, risk factors for prostate cancer are unknown (7). There are compelling reasons to believe that hormones, especially androgens, play a central role in prostate carcinogenesis. The growth and maintenance of the prostate are dependent on androgens (4), prostate cancer regresses following ablative or antiandrogen therapy (5), and testosterone can induce prostate cancer in laboratory rodents (6). However, epidemiologic data addressing hormonal hypotheses are limited and conflicting (7). In this issue of the Journal, Gann et al. (8) report their findings from a nested case-control study, using sera collected prior to prostate cancer diagnosis from a cohort of male physicians. In this study, higher levels of circulating testosterone and lower levels of estradiol and sex hormone-binding globulin (SHBG), a protein that binds about 56% of the circulating testosterone (9), are associated with increased risks of prostate cancer, while no association is found with DHT (the principal androgen within the prostate) or 3-alpha androstanediol glucuronide (a terminal metabolite of DHT that is used as an indirect measure of intraprostatic 5-alpha reductase activity) (9). Prospective nested case-control studies of this type provide better data for the tests of hormonal hypotheses than do casecontrol studies. Because blood was collected before the diagnosis of cancer, the possible impact of disease on hormonal levels is less profound and a temporal relationship can be established. To date, there are only five studies in which prediagnostic sera were used to examine the relationships of circulating hormones with subsequent prostate cancer risk (8,10-13). Results from these studies are inconsistent; positive associations were reported with androstenedione in one study, with testosterone-DHT ratio in two, and (weakly) with luteinizing hormone in one. Of these, Gann et al. is the largest and the first study to evaluate the role of 5-alpha reductase activity, using plasma levels of 3-alpha androstanediol glucuronide as a proxy measure. Although Gann et al. provide data consistent with a role for testosterone, a recent study by Nomura et al. (70), using a similar approach in a Japanese-American cohort in Hawaii, did not confirm this finding. The inconsistent results in these two well-designed studies may reflect, in part, the difficulties of measuring serum/plasma hormones in epidemiologic investigations. Sensitivity, specificity, and reliability of hormonal assays developed for clinical use need to be improved before they can be used for epidemiologic studies. The numbers of study subjects in most nested case-control studies are relatively small (<100) and the observed case-control differences are less than 15%, so that any potential noise, such as intra- and inter-person and laboratory variation, may influence hormonal assessment and case-control comparisons, leading to inconsistenci es between studies (14). To better interpret the case-control results, we need additional data on the reproducibility of hormonal assays, interrelationships among hormones, and factors influencing circulating levels of hormones.