The Impact of Common Genetic Variations in Genes of the Sex Hormone Metabolic Pathways on Steroid Hormone Levels and Prostate Cancer Aggressiveness

Sex steroid hormones influence the development, maturation, and maintenance of the prostate, affecting both the proliferation and differentiation status of the luminal epithelium (1–3). Androgens are synthesized primarily in the testes or adrenal glands, and secondarily in peripheral tissues including the prostate (2). Testosterone is the principal androgen present in the blood, mostly bound to sex hormone–binding globulin (SHBG) and albumin. Intracellularly, testosterone is enzymatically converted to the more potent metabolite dihydrotestosterone (DHT). The complex of androgen (testosterone or DHT) and androgen receptor regulates the expression of a variety of genes that are involved in the growth, survival, and differentiation of prostate cells by binding to androgen response elements present in the regulatory regions of these genes (2).

The relationship between androgens and the development of prostate cancer is complex. Some early studies showed that plasma testosterone levels might be correlated with Gleason grade or time to progression after orchiectomy of prostate cancer (3, 4). However, other studies indicated no evidence of any association between plasma testosterone, estradiol (E2), and SHBG with Gleason score at the time of diagnosis (5, 6). Thus, there is insufficient evidence to support the hypothesis that higher blood levels of steroid hormones are associated with prostate cancer development or progression. One explanation for this lack of association is the uncertain correlation between the plasma concentrations of androgens with the intraprostatic hormone levels, an essential component in the tumor microenvironment that potentially affects tumor growth (2, 7, 8). Another explanation is that a single determination of plasma androgens in middle or older age may not be representative of time-averaged or maximum levels over the etiologically relevant time of life (9). Similarly, the relationship between estrogen levels and prostate cancer is unclear. There is some evidence that estrogen metabolites functioning as genotoxins can increase risk for prostate cancer (10). It also has been suggested that a decrease in the androgen/estrogen ratio with aging could be responsible in part for prostate carcinogenesis (10).

Therapeutically, blocking androgen production has been one of the most commonly used treatments of advanced prostate cancer and is often successful until castration resistant growth is acquired (2, 11, 12). In addition, estrogens are recognized therapies for prostate cancer. Oral E2 administration may be useful for treatment of advanced castration-resistant prostate cancer (10). This therapy also has side effects and that the effects of estrogens are mediated by 2 different receptors, estrogen receptor α (ERα) and ERβ (10).

Genetic polymorphisms of genes involved in metabolic pathways of sex steroid hormones have been investigated as candidates for prostate cancer risk in many association studies (13–16). Data from these case–control studies suggest that interindividual variation in genes involved in sex hormone metabolic pathway could be related to the risk of prostate cancer incidence (13–16). However, few studies have examined the association of these polymorphisms with blood hormone levels in prostate cancer (17–19). In the National Cancer Institute (NCI)-Breast and Prostate Cancer Cohort Consortium (BPC3) study, single-nucleotide polymorphisms (SNP) in SHBG gene were found to be associated with circulating SHBG and testosterone levels in Caucasian men in both cases and controls. They also identified relating SNPs in ESR1 and SRD5A2 with testosterone and 3α-androstanediol-glucuronide, respectively (19).

We present here a systematic analysis of 143 common polymorphisms in 30 genes that are involved in the sex steroid hormones metabolic pathways for their potential associations with plasma total testosterone, total estradiol, and SHBG, or with prostate cancer aggressiveness in an established cohort of 508 patients with newly diagnosed localized prostate cancer.

The study cohort and the measurement of hormone levels have been previously described in detail (5, 20). Briefly, a total of 539 patients with localized prostate cancer evaluated at the Dana-Farber Cancer Institute from 2001 to 2005, who were enrolled in a prospective sample banking protocol, were eligible for this study. Most clinical data were abstracted from the Current Research Information System database, as described (21). Because mixed ethnicity could confound the association results, we removed 31 patients of non-Caucasian race or with indeterminate ancestry from this study. A total of 508 individuals were analyzed. These individuals either self-reported as European American or were determined to be of European ancestry, based on genotyping of ancestry informative markers. Using the D'Amico risk classification criteria, newly diagnosed patients with prostate cancer were classified as at low, intermediate, or high risk of clinical recurrence after primary therapy (22). Briefly, 3 risk groups were established on the basis of serum prostate-specific antigen (PSA) level, biopsy Gleason score, and clinical stage at diagnosis as described elsewhere (22). All the blood samples were drawn at around the time of diagnosis. Stored plasma was processed for the measurement of total testosterone, total estradiol, and SHBG levels by ELISA, as previously described (5).

Candidate genes were identified from review of the sex hormone biosynthesis and metabolic pathway (23–25) and cross-referenced with the Gene Ontology database to confirm pathway information. The 30 candidate genes that are included in this study and their roles in sex hormone synthesis and metabolic pathway are shown in Fig. 1. SNPs that were selected from each gene are of a minor allele frequency that was greater than 0.05 in the CEU population. The CEU HapMap population phase II genotyping data were used to select tagging SNPs. We ran the program Tagger and selected tags to capture the unmeasured variants r² more than 0.8. All nonsynonymous SNPs with a minor allele frequency more than 0.02 were also included (Table 2 and Supplementary Table S1). One hundred sixty-one SNPs in 30 genes were initially selected for analysis on the basis of their known participation in hormone synthesis and metabolism. All selected SNPs were included in the design. After genotyping, SNPs that did not conform to Hardy–Weinberg equilibrium (P < 0.01) or had genotyping call rates below 80% were omitted from further analysis (N = 18). Thus, a total of 143 SNPs in 30 genes were included for final statistical analyses. For quality control, about 5% random selected duplicates were included. Discrepancy rate was less than 0.1% between duplicates in overall genotyping data.

Descriptive statistics were used to characterize the study population in terms of frequency and percentage, or median and interquartile range. Plasma hormone levels were evaluated as continuous and D'Amico risk was dichotomized as low versus intermediate/high, the latter denoting aggressive disease. The Pearson χ² test and the Wilcoxon rank-sum test were used to evaluate global associations of SNP genotype with hormone levels and D'Amico risk, respectively. Relative risk (RR) and 95% confidence intervals (CI) were estimated by a generalized linear model for binomial data with a log-link rather than a logit-link function. The 2-sided P values of SNPs that were considered as statistically significant were based on an α of less than 0.05. Statistical analyses were conducted with the SAS software version 9.1 (SAS Institute Inc.).

As we were testing 143 different SNPs, we reported Q values, a measure of significance in terms of the false discovery rate (FDR) as proposed by Storey and Tibshirani for multiple comparisons (26). The Q value represents the expected proportion of false-positive results when announcing a test significant. Q values were estimated with R q-value package on the basis of the observed distribution of P values from the global association test for 143 individual SNPs. We considered a Q value less than 25% to be significant.

The characteristics of the current study cohort are shown in Table 1. In brief, patients had low- or intermediate-risk, early-stage, localized disease. The median PSA at diagnosis is 5.1 ng/dL. Seventy-nine percent of patients were diagnosed with T0 or T1 stage tumor. Fifty-three percent of patients' tumor Gleason score was 6 or less. In this cohort, 492 patients had sufficient information for criteria of D'Amico risk classification; and among them, 233 (47%) patients were at low risk, 202 (41%) intermediate risk, and 57 (12%) high risk. The majority of patients (82%) were overweight (body mass index BMI > 25 at the time of diagnosis). The average plasma total testosterone in this cohort was 440 ng/dL, the average plasma level of total estradiol was 9.2 ng/dL, and the average plasma SHBG level was 51 nmol/L. The average total estradiol to total testosterone ratio was 2.1.

We analyzed the association of plasma hormone levels and disease aggressiveness by D'Amico criteria with 143 SNPs from 30 genes that catalyze androgens and estrogens synthesis and bioactivation, catalyze the synthesis of catecholestrogens, and that metabolically inactivate androgens and estrogens (Fig. 1 and Table 2).

Among the 143 SNPs, 3 SNPs in HSD17B2 (rs1424151), HSD17B3 (rs9409407), and HSD17B1 (rs12602084), respectively, were found to be significantly associated with the level of plasma total testosterone (Table 3). The median total testosterone level in patients of the rs1424151 AG or GG genotype in HSD17B2 (371.6 ng/dL) was 17.4% less (P = 0.002) than in patients with the rs1424151 AA genotype (448.9 ng/dL). Similar results were observed for intronic SNP rs9409407 in HSD17B3, rs9409407 GT, or total testosterone genotype carriers having approximately 19.2% decrease in their median total testosterone (361.5 ng/dL) compared with the rs9409407 GG genotype carriers (447.6 ng/dL; P = 0.006). Individuals homozygous for rs12602084 in HSD17B1 gene had a median total testosterone (383.0 ng/dL), which is 14.5% less (P = 0.006) than those in carriers of the rs12602084 CC genotype (448.8 ng/dL) or carriers of the rs12602084 CT genotype (447.0 ng/dL). The Q values for all 3 SNPs were approximately 0.133.

We also analyzed the association between the 143 SNPs and total estradiol (Supplementary Table S1) or SHBG (data not shown) plasma levels. No significant association was observed.

We correlated the SNP genotyping data with the D'Amico risk categories at diagnosis in the study cohort. Of the 143 SNPs analyzed, 17 SNPs (12%) had an association with the risk of having an intermediate or high D'Amico risk class at diagnosis, with an unadjusted P value less than 0.05. Among them, 3 SNPs had Q values which were 0.248 and also had clear genotype–dose effect (Table 4). Individuals who carried the minor allele rs11252845 (CT or total testosterone genotype), a variant that is located in the upstream region of AKR1C1 gene, were found to have a reduced risk of having a intermediate or high D'Amico risk class at the time of diagnosis compared with the rs11252845 wild genotype (CC genotype) carriers (RR = 0.62; 95% CI, 0.40–0.88; P = 0.005). Similarly, individuals with the AT or AA genotype of rs2045100 in the first intron of UGT2B15 had an approximately 50% reduced risk of being diagnosed with a high-risk prostate cancer (RR = 0.54; 95% CI, 0.28–0.87; P = 0.007) compared with rs2045100 total testosterone genotype carriers. We also found that individuals who were homozygote for the minor allele genotype (GG genotype) of the intronic SNP, rs7932905, in HSD17B12, had a decreased risk of being diagnosed with a prostate cancer of intermediate or high D'Amico risk class (RR = 0.65; 95% CI, 0.48–0.86; P = 0.008) when compared with individuals carrying the rs7932905 AA genotype. Individuals carrying the rs7932905-AG genotype did not show a higher likelihood of having higher risk cancer at the time of diagnosis (RR = 0.89; 95% CI, 0.74–1.07).

We previously showed in this cohort the correlation of hormone levels with age and BMI (5). Weak associations were found between SHBG and BMI and age, and no relationship was observed among BMI, age, and sex steroid hormone levels (5). Prostate cancer aggressiveness–related inherited variants have been studied in several genetic association studies but have not been studied in sex steroid hormones metabolic pathway genes (27–30). Several association studies also tried to identify quantitative trait loci predicting sex steroid levels in men includes one recent large cohort study (17–19). The current study sought to determine whether polymorphisms in genes involved in steroid hormone metabolism may impact plasma steroid hormone levels at the time of diagnosis in patients with prostate cancer and the D'Amico risk categories of prostate cancer at diagnosis.

In this study, we observed that 3 noncoding SNPs, rs1424151 in HSD17B2, rs9409407 in HSD17B3, and rs12602084 in HSD17B1, were associated with significantly reduced plasma total testosterone level in newly diagnosed patients with prostate cancer. Intriguingly, some investigators have associated prostate cancer grade with low serum testosterone risk (31).

The human HSD17B gene family encodes the 17β-hydroxysteroid dehydrogenases that play a pivotal role in the production of steroid hormones (32, 33). HSD17B1 selectively catalyzes the transformation of androestenedione to testosterone and estrone (E1) to E2. HSD17B2 catalyzes the conversion of testosterone into androestenedione, converts 5-diol into DHEA, and converts E2 into E1 (34, 35). HSD17B3 plays an important role in the conversion from androstenedione to testosterone (35). Some studies indicated that changes of HSD17B1, HSD17B2, or HSD17B3 activity resulted in changes of the androgen level (33, 35). It is possible that these genetic polymorphisms in HSD17B1, HSD17B2, or HSD17B3 affect the expression of their encoded enzymes, consequently affecting plasma androgen level. This speculation remains to be determined.

We identified 3 SNPs related to circulating testosterone that differed from the NCI-BPC3 (19). Explanations for such discrepancies could include variability in sample size, diagnostic criteria, or ethnic background of the sample studied. In comparing the current study with the NCI-BPC3, there are a few such issues worth notice. First, the average age and BMI of the 2 populations are quite different. The median age of our cohort is 60 years, whereas the median age of NCI-BPC3 cohort is approximately 65 years. In our study, the majority (86%) of patients have a BMI of more than 25, whereas in NCI-BPC3 cohort only 58% had a BMI of more than 25 (19). Indeed, many studies have shown that age and BMI are tightly associated with circulating sex steroid levels (36, 37). Second, although case–control status in NCI-BPC3 study does not seem to be related to steroid levels, there was heterogeneity in measured sex steroid levels between different subcohorts, maybe due to the use of multiple laboratories. Our cohort consisted of only newly diagnosed early stage patients with prostate cancer, with a homogenous European ancestry. All our hormones levels were carried out in the same laboratory. Nonetheless, our findings need to be validated in other independent cohorts.

We did not observe any impact of these 143 SNPs on of the plasma total estradiol level. Because our subjects are patients with prostate cancer and total estradiol level in men is relatively low, any potentially subtle changes of total estradiol level in men, caused by the genetic polymorphisms in hormone metabolic pathway genes, may be difficult to identify.

This study of SNPs in metabolic pathway genes and prostate cancer aggressiveness identified 3 tagging SNPs, rs11252845 in ARK1C1, rs2045100 in UGT2B15, and rs7932905 in HSD17B12, borderline significantly associated with patients with prostate cancer D'Amico risk categories. Interestingly, both AKR1C1 and UGT2B15, involved in catabolic processes, were previously found highly expressed in androgen-independent castration-resistant prostate cancer tissues (38). AKR1C1 facilitates the conversion of DHT to 3α-androstanediol or 3β-androstanediol, a possible endogenous ligand for ERβ in prostate (38). UGT2B15, in conjunction with UGT2B17, mediates glucuronidation of DHT metabolites (40). HSD17B12 is a multifunctional isoenzyme functioning in the conversion of E1 to E2 and elongation of long-chain fatty acids, in particular the conversion of palmitic to archadonic acid, the precursor of sterols and the inflammatory mediator prostaglandin E. Its overexpression together with that of COX-2 in breast carcinoma is associated with a poor prognosis (41). The rs11252845 is located in the 5′ flanking region of the AKR1C1 and may potentially be involved in the regulation of AKR1C1 expression. The rs2045100 and rs7932905 are intronic SNPs. These 3 SNPs did not affect plasma levels of total testosterone or total estradiol. The possibility that they may influence intratumoral androgen is not excluded and remains to be determined.

In summary, of the 143 polymorphisms in the 30 genes involved in the steroid hormone metabolic pathways, we identified SNPs in 3 genes (HSD17B1, HSD17B2, and HSD17B3) that were associated with plasma total testosterone level, whereas SNPs in another 3 genes (AKR1C1, UGT2B15, and HSD17B12) might correlate with the risk of aggressive prostate cancer at diagnosis. Our study did not find SNPs that are significantly associated with both plasma hormone levels and prostate cancer aggressiveness. These results are consistent with our previous work finding no association between plasma steroid hormone levels and prostate cancer tumor grades.

No potential conflicts of interest were disclosed.

This work was supported by a SPORE in Prostate Cancer 2 P50 CA090381-06 and a Prostate Cancer Foundation Challenge grant. T. Sun is supported by a DoD Prostate Cancer Training Award W81XWH-09-1-0372.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.