by Pär Stattin, Sonja Lumme, Leena Tenkanen, Henrik Alfthan, Egil Jellum, Göran Hallmans, Steinar Thoresen, Timo Hakulinen, Tapio Luostarinen, Matti Lehtinen, Joakim Dillner, Ulf-Håkan Stenman, Matti Hakama

Androgens stimulate prostate cancer in vitro and in vivo. However, evidence from epidemiologic studies of an association between circulating levels of androgens and prostate cancer risk has been inconsistent. We investigated the association of serum levels of testosterone, the principal androgen in circulation, and sex hormone-binding globulin (SHBG) with risk in a case-control study nested in cohorts in Finland, Norway and Sweden of 708 men who were diagnosed with prostate cancer after blood collection and among 2,242 men who were not. In conditional logistic regression analyses, modest but significant decreases in risk were seen for increasing levels of total testosterone down to odds ratio for top vs. bottom quintile of 0.80 (95% CI = 0.59–1.06; ptrend = 0.05); for SHBG, the corresponding odds ratio was 0.76 (95% CI = 0.57–1.01; ptrend = 0.07). For free testosterone, calculated from total testosterone and SHBG, a bell-shaped risk pattern was seen with a decrease in odds ratio for top vs. bottom quintile of 0.82 (95% CI = 0.60–1.14; ptrend = 0.44). No support was found for the hypothesis that high levels of circulating androgens within a physiologic range stimulate development and growth of prostate cancer.

The hypothesis that androgens play a role in the pathogenesis of prostate cancer in men is mainly based on a large body of evidence from studies of tumor models.(1) In vitro, androgen response is observed in most well-differentiated cancer cell lines of prostatic origin,(2, 3) and in vivo, androgens consistently stimulate induction and promotion of prostate tumors and prostate tumor xenografts in rodent models, whereas androgen ablation causes tumor regression.(4, 5) In humans, an increased risk of prostate cancer may depend on an increase in testicular production of testosterone, resulting in high levels of circulating androgens, and/or tissue-specific alterations within the prostate increasing androgen stimulation only in the prostate.(1) At least 10 modest-sized prospective studies have investigated the former relationship, i.e., the association between circulating levels of androgens with prostate cancer risk.(6, 7, 8) Overall, the results from these studies have been inconclusive. Some studies have shown a mildly increased risk,(9, 10, 11, 12) whereas other studies have demonstrated a mildly decreased risk,(13, 14) but none of them have shown a statistically significant association between absolute levels of circulating testosterone, the principal androgen in the circulation, and prostate cancer risk. The biologically most active form of testosterone is considered to be bioavailable or free testosterone. Bioavailable testosterone is the fraction of testosterone that is not bound to sex hormone-binding globulin (SHBG), the dominant binding protein for testosterone in circulation; free testosterone is the fraction not bound to any protein. Given that albumin levels are constant between individuals, these 2 measures will be highly correlated. A few studies have investigated the association between serum levels of free testosterone and risk of prostate cancer. No significant association was observed in any of the 3 studies in which free testosterone was directly measured.(12, 14, 15) However, a significant increase was observed in one study in which bioavailable testosterone was calculated by linear regression adjustment of total testosterone for SHBG.10 Finally, several studies have shown a moderate decrease in risk for high levels of serum SHBG.(10, 13, 14, 15)

Serum levels of testosterone and SHBG are inversely correlated to obesity, which has been inconsistently associated with a weak increase in risk of prostate cancer.(7, 16, 17) We describe the results of a pooled case-control study nested in 3 prospective Nordic cohorts on the association of total and free testosterone and SHBG with prostate cancer risk, also taking obesity measured by body mass index into account.

Materials and Methods

Cohorts

We used data from cohorts of more than 200,000 men included in the Nordic Biological Specimen Biobank Working Group on Cancer, Causes and Control. The cohorts and the study design have previously been described in detail.18 In brief, the Janus project in Norway contains serum samples stored at −25°C from about 160,000 men who participated in county health examinations, mostly for cardiovascular diseases, and from blood donors. In Finland, the cohort consisted of approximately 19,000 men in the Helsinki Heart Study, a clinical drug trial. Serum from a blood sample was drawn at first screening visit and stored at −20°C. In Sweden, about 30,000 men have been recruited to the Northern Sweden Health and Disease Cohort as a representative population sample from the counties of Västerbotten and Norrbotten and plasma samples were stored at −80°C. For the subjects in the Finnish and Swedish cohorts, measurements of body mass index (BMI = weight/height2) were available. The study was approved by each respective local research ethics committee.

Case ascertainment and control selection

All incident cases of prostate cancer and all cases of death in the cohorts were identified through linkage with national or regional cancer and mortality registries. The local tumor stage according to Union International Contre le Cancer (UICC) was used.19 The study design was that 4 controls were randomly selected for each case from sets of all subjects alive and free of cancer at the time of diagnosis of the case and matching the case on cohort, age (± 2 year), county of residence and date (± 2 months in the Finnish and Swedish cohorts, ± 6 months in the Norwegian cohort) of blood sampling. Samples from 53 cases and 202 controls in the Finnish cohort had accidentally been thawed and were excluded from analysis.

Biochemical assays

The samples were analyzed in the Department of Clinical Chemistry, Helsinki University Hospital. The Finnish and Norwegian samples consisted of serum and the Swedish samples consisted of heparin plasma. Testosterone was quantitated by a time-resolved fluoroimmunoassay (Delfia, Wallac, Turku, Finland). The detection limit of the assay was 0.3 nmol/L. Intra- and interassay coefficients of variation at 2.4 nmol/L were 6.0% and 12.9%, at 10 nmol/L 5.5% and 6.8%, at 28 nmol/L 5.6% and 5.5%, respectively. SHBG was quantitated by a time-resolved immunofluorometric assay (AutoDelfia, Wallac). Samples were diluted 1:100 prior to assay. The detection limit of the assay was 0.5 nmol/L. Intra- and interassay coefficients of variation at 20 nmol/L were 1.4% and 8.2%, at 57 nmol/L 1.3% and 5.3%, at 13 nmol/L 1.8% and 10.1%, respectively.

Samples pertaining to matched study subjects were always analyzed together in the same batch (i.e., on the same day and within the same run). The laboratory personnel was blinded to the case-control status of the samples that was unmasked by the data manager after all biochemical analyses had been performed.

An index of free testosterone was calculated using a mass action equation based on the affinity of SHBG and albumin for testosterone, assuming an approximately constant serum albumin concentration for all men,(20) and such an index has been shown to be very highly correlated with free testosterone measured by a reference assay (r = 0.81 and 0.99).(21) We also estimated the free testosterone index as residuals of a linear regression of testosterone on levels of SHBG as done by Gann et al.(10)

Statistical analyses

Spearman coefficients of correlation were used to examine the cross-sectional relationships between serum hormone levels and BMI. A paired Student's t-test was used to test for differences between hormone levels of the cases and the mean values of their matched controls. Conditional logistic regression models were used to calculate odds ratios (ORs) for disease by quintiles of testosterone, SHBG and BMI. Cut points for quintiles were determined on variable distributions of cases and controls combined. The corresponding 95% confidence intervals were calculated with the Wald method. Likelihood ratio tests for linear trends in risk were performed on the variables categorized in quintiles using the score 1 through 5 for increasing hormone concentrations. To evaluate the nonlinear relationships between risk and exposure, the association between unordered categories of exposure and risk were assessed. To assess heterogeneity of ORs between cohorts, we compared logistic regression models with the main effects with and without interaction terms. Evidence for heterogeneity was regarded to exist if the difference between these models was significant. All statistical tests and corresponding p-values were 2-sided, and statistical analyses were performed with the SAS software package version 8.1 (SAS Institute, Cary, NC).

Results

There was a wide range in sample storage time, age and lag time between blood collection and diagnosis as an effect of the different designs and inclusion protocols in the 3 cohorts and these factors were closely related (Table I). The mean age at the time of blood collection for cases and controls was 46.6 years (range, 34–65) and 46.5 years (range, 34–64) in the Norwegian cohort, 51.2 years (range, 40–57) and 51.0 years (range, 40–57) in the Finnish cohort and 58.9 years (range, 40–73) and 59.0 years (range, 40–74) in the Swedish cohort. The mean lag time between blood collection and diagnosis was 16.7 years in the Norwegian cohort, 10.8 in the Finnish and 3.5 years in the Swedish cohort. The time from blood collection to hormonal analysis was 24.1 years in the Norwegian cohort, 17.6 in the Finnish and 7.0 years in the Swedish cohort.

The mean serum levels for cases compared to controls were lower for total testosterone (p = 0.01), SHBG (p = 0.01) and free testosterone (p = 0.28; Table II). In cross-sectional analyses, correlations between hormone levels were virtually identical for cases and controls; therefore, the correlations for cases and controls combined are reported. The indirect indexes for free testosterone calculated by a mass action equation and by linear regression of testosterone on levels of SHBG correlated almost perfectly (r = 0.99). There were strong direct correlations between total and free testosterone (r = 0.78) and between total testosterone and SHBG (r = 0.51), whereas the correlation between SHBG and free testosterone was very weak (r = −0.08). BMI was inversely correlated to SHBG (r = −0.33) and total testosterone (r = −0.28), and less strongly to free testosterone (r = −0.14). Mean concentrations of hormones for increasing levels of BMI are shown in Figure 1.

Table II. Mean Serum Levels of Free and Total Testosterone, SGBG and BMI by Case Control Status and Country

Figure 1.
Means of hormonal levels by quintile levels of BMI.

We found mild, albeit significant, decreases in risk for increasing levels of total testosterone down to an OR for top vs. bottom quintile of 0.80 (95% CI = 0.59–1.06; ptrend = 0.05) and the corresponding OR for SHBG was 0.76 (95% CI = 0.57–1.01; ptrend = 0.07; Fig. 2a and b). Separate analyses of the 3 cohorts showed the same essentially linear risk pattern as in the full study group. Mutual adjustments for SHBG and total testosterone only very slightly attenuated the decreases in risk. ORs for testosterone adjusted for SHBG were 1.00 (ref), 1.07 (95% CI = 0.81–1.40), 1.23 (95% CI = 0.94–1.62), 0.90 (95% CI = 0.66–1.22) and 0.93 (95% CI = 0.66–1.31). For free testosterone, a bell-shaped risk pattern was seen, with an OR of 0.82 (95% CI = 0.60–1.14) for top vs. bottom quintile (ptrend = 0.44; Fig. 2c), and an OR of 0.63 (95% CI = 0.47–0.85) for top vs. middle quintile. In the Finnish cohort, OR for top vs. bottom quintile of free testosterone was increased to 1.40 (95% CI = 0.59–3.31), but essentially the same risk pattern was seen in the Norwegian and Swedish cohorts as in the full cohort. There was a significant nonlinear association between free testosterone and risk in the full study group (p = 0.02) and in the Norwegian cohort (p = 0.05), but not in the Finnish and Swedish cohorts separately or combined (p = 0.52). No significant heterogeneity was found between cohorts in the effect of hormone levels on risk in formal tests (data not shown).

ORs for increasing quintiles of BMI, available only in the Finnish and Swedish study cohorts, were 1.00 (ref), 1.62 (95% CI = 0.94–2.80), 1.05 (95% CI = 0.60–1.84), 1.05 (95% CI = 0.60–1.85) and 1.41 (95% CI = 0.83–2.40; ptrend = 0.68). Risk estimates for hormonal measurements did not change materially when adjusted for BMI. After adjustment for BMI, OR for top vs. bottom quintile changed for total testosterone from 1.01 (95% CI = 0.56–1.81) to 1.08 (95% CI = 0.58–2.00), for SHBG from 0.62 (95% CI = 0.36–1.08) to 0.66 (95% CI = 0.36–1.19) and for free testosterone from 1.10 (95% CI = 0.58–2.07) to 1.05 (95% CI = 0.60–2.20). Similarly, the risk estimates remained essentially unchanged when analyses were done in subgroups according to age and lag time (data not shown).

We investigated the risk associated with extreme levels of hormones on prostate cancer risk and found virtually the same patterns of risk as for top vs. bottom quartiles. ORs for bottom and top deciles (10%) vs. all other deciles combined were for total testosterone 0.89 (95% CI = 0.66–1.20) and 0.84 (95% CI = 0.62–1.14), for SHBG 1.07 (95% CI = 0.80–1.43) and 0.67 (95% CI = 0.49–0.93) and for free testosterone 0.68 (95% CI = 0.49–0.94) and 0.74 (95% CI = 0.54–1.02), respectively.

Discussions

In this pooled prospective study, 3 times larger than any previous study, we observed a modest but significant decrease in prostate cancer risk for increasing levels of total testosterone. For SHBG, a near-significant decrease was seen; for free testosterone, a bell-shaped risk pattern with a nonsignificant decrease for the highest level.

With regard to levels of androgen exposure, we found an approximately 2-fold difference in serum testosterone in the highest vs. the lowest exposure groups in accordance with previous studies.(6) An equally large increase in androgen administration has produced substantial increases in induction and promotion of prostate tumors in rodent tumor models.1 However, short-term administration of androgens in rodent models may not be equivalent to chronic exposure to mildly elevated endogenous androgen levels in men.

Most sex steroids and SHBG are fairly robust in cryopreserved samples even for relatively long periods,(22, 23) but some degradation cannot be ruled out in the samples that had been cryopreserved for a very long time at a less than ideal temperature. However, as the cases and controls were closely matched for date of blood collection and also for cohort, degradation would affect samples from cases and controls nondifferentially and would only tend to attenuate an association to risk. Matching for cohort should obviate any possible effect of different sample type, which was serum for Finnish and Norwegian cohorts and plasma in the Swedish cohort. Testosterone levels but not SHBG levels were substantially higher in Swedish cases and controls than in subjects in the other cohorts despite the higher mean age in the Swedish cohort. We speculate that this difference in concentrations may have been due to a shorter storage time and lower temperature in the Swedish biorepository. Serum levels of testosterone and SHBG are relatively constant over time in an individual, with reported correlation coefficients ranging from 0.7 to 0.9 between levels in samples taken about 3 years apart.(23, 24, 25, 26) However, if one single blood sample is representative for exposure to hormonal levels over longer time periods is unknown. There was a wide range in age at exposure, storage time and lag time in our study, but as the Norwegian cohort was more than twice the size of the Finnish and Swedish cohorts combined, it largely dominated the full cohort analysis. Subjects in the Norwegian cohort were much younger at recruitment and time between recruitment and diagnosis as well as storage times were consequently substantially longer in the Norwegian cohort than in the other 2 cohorts.

The mean lag time in our study was 14 years, which is much longer than in all but 2 small previous studies.6, 27 Given that the average preclinical duration of prostate cancer has been calculated to more than a decade,(28, 29) the majority of blood samples in our study reflected the hormonal exposure during an earlier stage of tumorigenesis than in most previous studies. However, the risk estimates in the Swedish and Finnish cohorts were similar to those obtained in the Norwegian cohort and we think that the similarity in risk estimate despite the differences in cohort characteristics adds further strength to our results.

Concentrations of free testosterone, biologically the most active form of testosterone, is most accurately directly measured by a dialysis method, which has been used rather infrequently in epidemiologic studies due to the requirements of large sample volumes and time-consuming methodology. However, such direct measurements have been shown to correlate well with indexes of free testosterone calculated from total testosterone and SHBG levels as used by us.(20, 21, 23)

Total testosterone linearly adjusted for SHBG can also be used as a calculated index of biologically active testosterone, and one often-cited report from the Physicians' Health Study showed a significant increase in prostate cancer risk for total testosterone linearly adjusted for SHBG.10 In contrast, we observed modest nonsignificant decreases in risk for testosterone adjusted for SHBG. Similarly, for an index of free testosterone calculated from total testosterone and SHBG, we observed a bell-shaped pattern of risk, with weakly increased risk for the intermediate level and decreased risk for the highest level. We have no obvious explanation for this risk pattern.

Indirect supports for the hypothesis that high levels of circulating androgens is a risk factor for prostate cancer have included the dramatic regression of tumor symptoms in a majority of men with advanced prostate cancer after androgen ablation, which entails a total cessation of testicular production of testosterone.30 However, the effects of the drastic reduction in androgen levels seen at a very late stage of cancer development may not be relevant to the effect of variations within a physiologic range on early tumor events that takes place decades earlier. Furthermore, the observation that men in populations with low prostate cancer incidence have lower levels of circulating androgens than men in populations with high incidence has not been consistent,(26, 31) and may in fact reflect mainly the differences in prevalence of obesity in different populations.(24)

Our finding of a mild, albeit significant, decrease in risk for total testosterone essentially concurs with data from a metaanalysis by Eaton et al.,(6) in which no significant association between testosterone and risk was found in an analysis of data from 8 prospective studies with a total number of 817 cases and 2,107 controls. Thus, taken together, data from all published prospective studies on circulating levels of total and free testosterone do not support the hypothesis that high levels of circulating androgens are associated with an increased risk of prostate cancer.(6, 7)

Total testosterone and SHBG, but not free testosterone, were significantly inversely correlated to BMI in our study, in accordance with previous reports.(26, 32, 33) Furthermore, BMI was nonsignificantly associated with increased risk, and globally, prospective studies have yielded somewhat inconsistent evidence of a weak direct association between obesity measured by BMI and prostate cancer risk.(7, 16, 17) We observed rather strong negative correlations between BMI, SHBG and total testosterone, but a less strong correlation between BMI and free testosterone, which fits with the observation that an obesity-induced decrease in SHBG levels decreases testosterone levels through a feedback mechanism of free testosterone on the hypothalamic-pituitary-gonadal axis.7 Our observation of a mild inverse association between SHBG and risk down to a relative risk of 0.76, consistent with most previous studies,(10, 13, 14, 15) is in agreement with obesity as a weak risk factor for prostate cancer. Thus, the fact that obesity is not a protective factor for prostate cancer supports the idea that circulating testosterone is not a major risk factor for prostate cancer.(7)

Circulating androgen levels can be important for prostate cancer development only if they accurately reflect intraprostatic androgen signaling and little is known if they do. Intraprostatatic androgen signaling may depend more on the rate of conversion of testosterone by the 5α-reductase II enzyme to dihydrotestosterone (DHT), the most potent androgen in the prostate, and on the propensity for androgen receptor activation than on circulating androgen levels.(1, 34, 35) Polymorphisms affecting 5α-reductase activity and pharmacologic blockade of 5α-reductase conversion of testosterone to DHT have been associated with prostate cancer risk in some studies.(34, 36, 37) Furthermore, androstanediol glucuronide (A-diol-g), the main circulating metabolite of DHT, was the only androgen that was significantly increased in prostate cancer cases in the previously cited metaanalysis.6 However, circulating A-diol-g levels only partially reflect intraprostatic DHT levels as A-diol-g is also produced by the 5α-reductase I enzyme present in the skin. Finally, variations in the length of CAG repeats in the androgen receptor affect DNA transcriptional activity and may also influence development of prostate cancer.(35, 38)

A clinical implication of our results concerns androgen supplementation, which has become easier to administer with the advent of transdermal patches.39, 40 Androgen supplementation has been suggested for treatment of mild hypogonadism, and reports have included beneficial effects on erectile dysfunction, abdominal obesity, osteoporosis, muscle strength and angina pectoris.(40, 41) There has been concern that androgen supplementation in the same order of magnitude as endogenous testicular production of testosterone (3–10 mg/24 hr) may promote prostate cancer but our results do not support this view. SHBG levels were slightly decreased or not affected by parenteral administration of testosterone in most studies and consequently levels of free testosterone have been proportionally more increased than levels of total testosterone.(39) Thus, it is uncertain whether the effect of variations within the physiologic range of endogenous testosterone levels can be inferred to the effect of a constant exogenous supply of testosterone, probably bypassing the normal coregulation of SHBG and testosterone.

Another implication of our results is that in a search for hormonal changes and genetic polymorphisms associated with prostate cancer risk, changes affecting androgen metabolism downstream of testosterone are the ones most likely to yield positive findings.

In conclusion, in this large pooled prospective study, we found mild, albeit significant, linear decreases in risk for increasing levels of total testosterone, linear near-significant decreases for SHBG and a bell-shaped risk pattern for free testosterone with a nonsignificant decrease for the highest level. Thus, no support was found for the hypothesis that circulating androgens stimulate prostate cancer development and growth.

Acknowledgements 

The Janus serum bank is owned and financed by the Norwegian Cancer Society. Randi Gislefoss and Nina Rokke serve as Janus coordinators and Åsa Ågren serves as coordinator for the Northern Sweden health and disease cohort. This is study no. 19 from the Nordic Biological Specimen Biobanks Working Group on Cancer, Causes and Control.

References

1. Bosland MC. The role of steroid hormones in prostate carcinogenesis. J Natl Cancer Inst 2000; 27: 67–73.
2 Kyprianou N, English HF, Isaacs JT. Programmed cell death during regression of PC-82 human prostate cancer following androgen ablation. Cancer Res 1990; 50: 3748–53.
3 Webber MM, Bello D, Quader S. Immortalized and tumorigenic adult human prostatic epithelial cell lines: characteristics and applications—1, cell markers and immortalized nontumorigenic cell lines. Prostate 1996; 29: 386–94.
4 Lucia MS, Bostwick DG, Bosland M, Cockett AT, Knapp DW, Leav I, Pollard M, Rinker-Schaeffer C, Shirai T, Watkins BA. Workgroup I: rodent models of prostate cancer. Prostate 1998; 36: 49–55.
5 Bladou F, Vessela RL, Buhler KR, Ellis WJ, True LD, Lange PH. Cell proliferation and apoptosis during prostatic tumor xenograft involution and regrowth after castration. Int J Cancer 1996; 67: 785–90.
6Eaton NE, Reeves GK, Appleby PN, Key TJ. Endogenous sex hormones and prostate cancer: a quantitative review of prospective studies. Br J Cancer 1999; 80: 930–4.
7 Kaaks R, Lukanova A, Sommersberg B. Plasma androgens, IGF-I, body size and prostate cancer risk: a synthetic review. Prostate Cancer Prostatic Dis 2000; 3: 157–72.
8 Hsing AW, Reichardt JK, Stanczyk FZ. Hormones and prostate cancer: current perspectives and future directions. Prostate 2002; 52: 213–35.
9 Hsing AW, Comstock GW. Serological precursors of cancer: serum hormones and risk of subsequent prostate cancer. Cancer Epidemiol Biomarkers Prev 1993; 2: 27–32.
10 Gann PH, Hennekens CH, Ma J, Longcope C, Stampfer MJ. Prospective study of sex hormone levels and risk of prostate cancer. J Natl Cancer Inst 1996; 88: 1118–26.
11 Heikkila R, Aho K, Heliovaara M, Hakama M, Marniemi J, Reunanen A, Knekt P. Serum testosterone and sex hormone-binding globulin concentrations and the risk of prostate carcinoma: a longitudinal study. Cancer 1999; 86: 312–5.
12 Guess HA, Friedman GD, Sadler MC, Stanczyk FZ, Vogelman JH, Imperato-McGinley J, Lobo RA, Orentreich N. 5 alpha-reductase activity and prostate cancer: a case-control study using stored sera. Cancer Epidemiol Biomarkers Prev 1997; 6: 21–4.
13 Vatten LJ, Ursin G, Ross RK, Stanczyk FZ, Lobo RA, Harvei S, Jellum E. Androgens in serum and the risk of prostate cancer: a nested case-control study from the Janus serum bank in Norway. Cancer Epidemiol Biomarkers Prev 1997; 6: 967–9.
14 Dorgan JF, Albanes D, Virtamo J, Heinonen OP, Chandler DW, Galmarini M, McShane LM, Barrett MJ, Tangrea J, Taylor PR. Relationships of serum androgens and estrogens to prostate cancer risk: results from a prospective study in Finland. Cancer Epidemiol Biomarkers Prev 1998; 7: 1069–74.
15 Nomura AM, Stemmermann GN, Chyou PH, Henderson BE, Stanczyk FZ. Serum androgens and prostate cancer. Cancer Epidemiol Biomarkers Prev 1996; 5: 621–5.
16 Bianchini F, Kaaks R, Vainio H. Overweight, obesity, and cancer risk. Lancet Oncol 2002; 3: 565.
17 Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of US adults. N Engl J Med 2003; 348: 1625–38.
18 Stattin P, Adlercreutz H, Tenkanen L, Jellum E, Lumme S, Hallmans G, Harvei S, Teppo L, Stumpf K, Luostarinen T, Lehtinen M, Dillner J, Hakama M. Circulating enterolactone and prostate cancer risk: a Nordic nested case-control study. Int J Cancer 2002; 99: 124–9.
19 Hermanek P, Sobin LH. TNM classification of malignant tumours: International Union Against Cancer. Berlin: Springer-Verlag, 1992. 141–4.
20 Anderson DC, Thorner MO, Fisher RA, Woodham JP, Goble HL, Besser GM. Effects of hormonal treatment on plasma unbound androgen levels in hirsute women. Acta Endocrinol 1975; 199 (Suppl): 224.
21 Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 1999; 84: 3666–72.
22 Sinnecker G. Stability of sex-hormone-binding globulin in serum and plasma. Clin Chem 1989; 35: 1253–4.
23 Morley JE, Patrick P, Perry HM III. Evaluation of assays available to measure free testosterone. Metabolism 2002; 51: 554–9.
24 Gapstur SM, Gann PH, Kopp P, Colangelo L, Longcope C, Liu K. Serum androgen concentrations in young men: a longitudinal analysis of associations with age, obesity, and race—the CARDIA male hormone study. Cancer Epidemiol Biomarkers Prev 2002; 11: 1041–7.
25 Vermeulen A, Verdonck G. Representativeness of a single point plasma testosterone level for the long term hormonal milieu in men. J Clin Endocrinol Metab 1992; 74: 939–42.
26 Ukkola O, Gagnon J, Rankinen T, Thompson PA, Hong Y, Leon AS, Rao DC, Skinner JS, Wilmore JH, Bouchard C. Age, body mass index, race and other determinants of steroid hormone variability: the HERITAGE family study. Eur J Endocrinol 2001; 145: 1–9.
27 Carter HB, Pearson JD, Metter EJ, Chan DW, Andres R, Fozard JL, Rosner W, Walsh PC. Longitudinal evaluation of serum androgen levels in men with and without prostate cancer. Prostate 1995; 27: 25–31.
28 Etzioni R, Cha R, Feuer EJ, Davidov O. Asymptomatic incidence and duration of prostate cancer. Am J Epidemiol 1998; 148: 775–85.
29 Stenman UH, Leinonen J, Zhang WM, Finne P. Prostate-specific antigen. Semin Cancer Biol 1999; 9: 83–93.
30 Nair B, Wilt T, MacDonald R, Rutks I. Early versus deferred androgen suppression in the treatment of advanced prostatic cancer. Cochrane Database System Rev 2002; 1: CD003506.
31 Pettaway CA. Racial differences in the androgen/androgen receptor pathway in prostate cancer. J Natl Med Assoc 1999; 91: 653–60.
32 Tibblin G, Adlerberth A, Lindstedt G, Bjorntorp P. The pituitary-gonadal axis and health in elderly men: a study of men born in 1913. Diabetes 1996; 45: 1605–9.
33 Stellato RK, Feldman HA, Hamdy O, Horton ES, McKinlay JB. Testosterone, sex hormone-binding globulin, and the development of type 2 diabetes in middle-aged men: prospective results from the Massachusetts male aging study. Diabetes Care 2000; 23: 490–4.
34 Irani J, Ravery V, Pariente JL, Chartier-Kastler E, Lechevallier E, Soulie M, Chautard D, Coloby P, Fontaine E, Bladou F, Desgrandchamps F, Haillot O. Effect of nonsteroidal anti-inflammatory agents and finasteride on prostate cancer risk. J Urol 2002; 168: 1985–8.
35 Giovannucci E, Stampfer MJ, Krithivas K, Brown M, Dahl D, Brufsky A, Talcott J, Hennekens CH, Kantoff PW. The CAG repeat within the androgen receptor gene and its relationship to prostate cancer. Proc Natl Acad Sci USA 1997; 94: 3320–3.
36 Norman RW, Coakes KE, Wright AS, Rittmaster RS. Androgen metabolism in men receiving finasteride before prostatectomy. J Urol 1993; 150: 1736–9.
37 Makridakis NM, Reichardt JK. Molecular epidemiology of hormone-metabolic loci in prostate cancer. Epidemiol Rev 2001; 23: 24–9.
38 Montgomery JS, Price DK, Figg WD. The androgen receptor gene and its influence on the development and progression of prostate cancer. J Pathol 2001; 195: 138–46.
39 Gruenewald DA, Matsumoto AM. Testosterone supplementation therapy for older men: potential benefits and risks. J Am Geriatr Soc 2003; 51: 101–15.
40 Vermeulen A. Androgen replacement therapy in the aging male: a critical evaluation. J Clin Endocrinol Metab 2001; 86: 2380–90.
41 Morales A. Androgen replacement therapy and prostate safety. Eur Urol 2002; 41: 113–20.