V89L Polymorphism of the 5α-Reductase Type II Gene (SRD5A2), Endogenous Sex Hormones, and Prostate Cancer Risk

Prostate cancer is the most common cancer and is the second most common cause of cancer mortality in American men (1). Androgenic hormones are essential for the growth and development of the prostate gland (2) and androgen-ablative therapies are commonly used to treat prostate cancer (3). Therefore, it has been hypothesized that high circulating androgen levels are associated with an increased risk of the disease. Previous studies that examined prediagnostic hormone levels and the risk of prostate cancer found mixed results. Some studies observed an increase in the risk of prostate cancer with increasing levels of plasma testosterone (4, 5). In others, increased risk of prostate cancer was associated with increasing levels of androstenedione (6), and with decreasing levels of estrogen (5). However, most of the associations were relatively weak. Other studies have found no associations between risk and hormone levels (7, 8).

The 5α-reductase II enzyme converts testosterone into dihydrotestosterone, a more potent natural androgen than testosterone. It has been hypothesized that an increase in the 5α-reductase enzyme activity is associated with an increased risk of prostate cancer. The gene that codes the 5α-reductase II enzyme, SRD5A2, has a G to C transversion resulting in a valine (V) for leucine (L) amino acid substitution at codon 89 (V89L polymorphism, rs523349). The L/L genotype tends to have decreased enzyme activity, and in some studies, was associated with lower androgen levels (9, 10), but not in others (11). The L/L genotype is more frequent among Asians, a group with a lower incidence of prostate cancer (12). Several studies have examined the possible association between the V89L genotype of the SRD5A2 gene and the risk of prostate cancer. Although some found a decrease in risk in L/L men compared with V/V men (13-15), others did not (10, 11, 16, 17).

To our knowledge, none of the studies to date have examined the combined effect of hormone levels and V89L polymorphism on prostate cancer risk. We hypothesized that men who exhibit high levels of androgens in addition to a more active 5α-reductase enzyme, as inferred from the SRD5A2 genotype, would be at increased risk of prostate cancer compared with men with low levels of androgens and low 5α-reductase activity. To test this hypothesis, we examined the risk of prostate cancer associated with above- and below-median levels of androstenedione, total testosterone and total estradiol, free testosterone and free estradiol (based on the idea that it is the free fraction that may exert an effect), 3α-diol glucuronide (3α-diol G), and dehydroepiandrosterone sulfate (DHEAS) among men with the V/V versus V/L or V/V genotypes.

We conducted a nested case control study of men who participated in the Carotene and Retinol Efficacy Trial (CARET), a lung cancer prevention trial testing β-carotene and retinol palmitate among individuals who are at high risk (18, 19). Two types of individuals were eligible for the CARET study: (a) Asbestos exposed. Men between the ages of 45 to 69 years who were current or former smokers with occupational exposure beginning at least 15 years prior to enrollment and were either employed in a CARET protocol–defined high-risk trade for at least 5 years or had a chest X-ray image compatible with asbestos-related lung disease. (b) Heavy smokers. Men or women between the ages of 50 to 69 who were current or recent smokers (quit <6 years before beginning of study) with a history of at least 20 pack-years of smoking. A total of 18,314 participants were enrolled to the CARET study and were followed from 1985 to 2003.

All CARET participants were followed periodically for cancer diagnosis. Cancer was confirmed by review of medical records and pathology report. A total of 300 CARET study participants who developed prostate cancer between 1987 and 1998 and who had not been diagnosed with lung cancer were chosen for our study. Men had available blood samples from at least 6 months prior to their diagnosis. A medical oncologist (G.E. Goodman) reviewed the medical records for staging and grading of the tumors. American Joint Committee on Cancer staging was based on pathologic information, if available, or on clinical staging information.

Controls were men who participated in the CARET study and were free of lung or prostate cancer as of August 1999. Controls were matched to cases on race, age at enrollment within 5 years, time of day of blood draw used for hormone assays (within 2 h), CARET study center, and year of randomization (randomized within a year of the case). Controls were required to have been alive and under follow-up at least until the date of diagnosis of the paired case. All the participants provided informed consent and all the institutional review offices of the participating centers approved the study.

Prediagnosis blood samples (cases and control) were analyzed in Chu Chen's Laboratory at the Fred Hutchinson Cancer Research Center. The laboratory analyses included: (a) Hormone profiles (described previously in ref. 8), including sex hormone binding globulin, androstenedione, testosterone, estradiol, 3α-diol glucuronide, and DHEAS. Intra-assay and inter-assay %CV for these assays in the Chen Laboratory were 3 to 5 and 5 to 6 for sex hormone binding globulin, 6 to 10 and 6 to 12 for androstenedione, 6 to 12 and 9 to 14 for testosterone, 7 to 13 and 12 to 18 for estradiol, 2 to 3 and 7 to 10 for 3α-diol G, and 1 to 2 and 2 to 4 for DHEAS, respectively. It has been shown that serum testosterone and dihydrotestosterone concentration do not correlate well with their respective prostate levels (20), and that measurements of the circulating levels of DHEAS and of 3α-diol G better reflect intraprostatic androgenicity (7, 21). Sex hormone binding globulin was used in the estimation of free testosterone and free estradiol levels. (b) V89L polymorphism determination of the gene for 5α-reductase type II (SRD5A2; described previously in ref. 17).

Logistic regression analysis for paired data with robust variance estimates was used to estimate the odds ratios (OR) for prostate cancer associated with the different genetic and hormonal profiles. The analysis included the following hormones: androstenedione, total and free testosterone, total and free estradiol, 3α-diol G, and DHEAS. For each of the hormones, two groups (above and below the median) were defined. V89L polymorphism was also divided into two groups, the V/V genotype versus the V/L or L/L genotype. Each analysis examined the OR for prostate cancer of individuals with high hormone concentrations versus low hormone concentrations with the same level of enzyme activity [a more active enzyme (V/V genotype) or a less active enzyme (V/L or L/L genotype)]. An interaction term between the hormone level and the SRD5A2 genotype was added. Low hormone concentrations were used as a reference. All analyses were adjusted for age and race and were not adjusted for the other hormone's concentrations.

Subanalyses stratified by aggressiveness of the disease were conducted in which aggressive cancers were defined as stage C or D, or as stage A or B but having a Gleason score of ≥7. The statistical analyses were done using STATA statistical software version 9.1 for windows.

Selected baseline characteristics of the study participants are summarized in Table 1. Cases and controls were similar in age, body mass index, marital status, and educational level. Both groups included ∼50% former smokers and 50% current smokers at baseline. The participants were 94% Caucasian, 3% African American, and 4% from other ethnic groups. Cases and controls had similar proportions of patients randomized into the placebo and treatment groups.

Baseline serum hormone levels of the study participants are summarized in Table 2. The cases and control were similar with regards to their baseline androstenedione, total and free testosterone, and total and free estradiol. The cases had a higher level of DHEAS and 3α-diol G.

The distribution of the V89L genotypes was similar for both cases and controls. Ten percent of the participants had an L/L genotype (29 of the cases, 30 of the controls), ∼40% had a V/L genotype (124 of the cases, 120 of the controls), and ∼50% of the participants had a V/V genotype (147 of the cases, 150 of the controls).

A moderate increase in risk associated with above-median serum levels of androstenedione and DHEAS was present irrespective of V89L genotype (Table 3). However, in L/L or V/L men, above-median DHEAS levels were associated with an increased risk of aggressive tumors (OR, 3.12; 95% CI, 1.28-7.63) but not of nonaggressive ones (OR, 0.56; 95% CI, 0.25-1.25).

Above-median serum levels of total testosterone tended to be associated with a lower risk of cancer in men with an L/L or V/L genotype, whereas above-median serum levels of free testosterone tended to be associated with a lower risk of cancer regardless of V89L genotype (Table 3).

There was a trend for above-median serum levels of total estradiol to be associated with a lower risk of cancer, especially nonaggressive cancer (Tables 3 and 4). The inverse association with nonaggressive cancer was observed regardless of genotype (OR, 0.34; 95% CI, 0.13-0.88 in men with an L/L or V/L genotype; OR, 0.39; 95% CI, 0.15-1.02 in men with a V/V genotype). There was an indication that above-median serum levels of free estradiol were associated with a lower risk of cancer, especially aggressive cancer (Tables 3 and 4). The inverse association with aggressive cancer was more pronounced in men with a V/V genotype (OR, 0.34; 95% CI, 0.14-0.81), than in men with an L/L or V/L genotype (OR, 0.77; 95% CI, 0.37-1.60).

Levels of 3α-diol G above the median were associated with an increased risk of prostate cancer, but only in men with the L/L or V/L genotype (OR, 2.16; 95% CI, 1.31-3.56; Table 3). Comparing the upper fourth of the 3α-diol G distribution to the lower half did not reveal any additional difference in risk (OR, 2.0; 95% CI, 1.15-3.49). Among men with the V/V genotype, above-median levels of 3α-diol G were not associated with an altered risk of prostate cancer (OR, 1.04; 95% CI, 0.65-1.66). The increase in risk associated with above-median levels of 3α-diol G in L/L and V/L men was restricted to aggressive tumors (Table 4).

We observed a positive association between above-median serum levels of DHEAS and 3α-diol G and the risk of aggressive prostate cancer. The association differed by the 5α-reductase type II genotype, being evident only in men with the L/L or V/L genotype. The inverse association between serum levels of free estradiol and aggressive prostate cancer was most prominent in men with the V/V genotype. The inverse association between serum levels of total estradiol and nonaggressive cancer was observed regardless of 5α-reductase type II genotype. To our knowledge, this is the first study that examined the association of prostate cancer and endogenous hormone levels stratified by V89L genotype.

A weak association of high serum levels of androstenedione with an increased risk of prostate cancer was reported in previous studies (6, 22), but the association between androstenedione levels and risk of prostate cancer in conjunction with SRD5A2 genotype has not been examined before. We found that above-median serum levels of androstenedione were weakly associated with an increase in the risk of cancer, regardless of V89L genotype (OR, 1.37; 95% CI, 0.85-2.21 for the L/L or V/L genotypes; OR, 1.25; 95% CI, 0.78-2.02 for the V/V genotype).

In two previous studies (8, 23), serum levels of DHEAS bore, at most, a weak association with risk of prostate cancer. In the present study, elevated DHEAS levels were observed to be associated with an increased risk of aggressive prostate cancer in men with L/L or V/L genotypes. DHEA can be converted to testosterone via conversion to either androstenedione or 5-androstenediol. It is possible that a high enough concentration of DHEA (as measured by DHEAS) could result in significant testosterone production, which results in a higher conversion rate to dihydrotestosterone, even in patients with low 5α-reductase activity.

There are conflicting findings regarding serum estradiol levels and prostate cancer. Although two studies observed an association of high levels of estradiol with a decrease in risk of prostate cancer (5, 8), others found a trend for high levels of estradiol to be associated with an increase in the risk of cancer (6), and some reported no association with prostate cancer at all (4, 23). We found that, adjusted for age and race, above-median levels of free estradiol were associated with a decrease in the risk of aggressive prostate cancer, especially in V/V men. However, this pattern of results was not observed in analyses based on levels of total estradiol rather than those of free estradiol.

Three prospective studies have found a weak trend for an increase in the risk of prostate cancer with elevated levels of 3α-diol G (5, 22, 24). Gann et al. and Nomura et al. looked at all stages of prostate cancer, whereas the study of Guess et al. was limited to cancers grade B or higher. In our parent study (8), a trend for an increase in the risk of prostate cancer with elevated levels of 3α-diol G was found, which was stronger for aggressive prostate cancer. In the current study, we found that adjusted for age and race, above-median levels of 3α-diol G were associated with an increase in the risk of prostate cancer in men with L/L or V/L genotype, more so for aggressive tumors. 3α-diol G is a metabolite of dihydrotestosterone. High levels of 3α-diol G in patients with low 5α-reductase can imply a possible association between prostate cancer and dihydrotestosterone metabolism that might warrant further investigation.

There are some limitations to our study. Prostate cancer can be a slow-growing tumor. Hormone levels were measured on blood specimens drawn 6 months to 5 years before diagnosis. Some of the cases probably had undiagnosed prostate cancer at this time, which conceivably could have influenced hormone levels. In addition, hormone levels measured later in life might not correlate well with levels earlier in life, and those earlier levels could be more relevant for the development of prostate cancer. Misclassification could also arise from having only a single blood sample on which to make the hormone measurements. Finally, our sample size was large enough to identify only moderate-to-large associations.

In conclusion, our study observed that only in men with the L/L or V/L genotype of the V89L polymorphism of the 5α-reductase type II gene were increased serum levels of DHEAS and 3α-diol G positively associated with a higher risk of aggressive prostate cancer. Free estradiol levels were negatively associated with risk of aggressive prostate cancer in men with the V/V genotype. However, the absence of an overall association between V89L genotype and aggressive prostate cancer (17) argues for a cautious interpretation of these observations, pending the availability of data from other studies.

We thank the CARET study participants for providing the demographic information and biological specimens, and the principal investigators and staff at the participating CARET study centers for specimens and data collection.