Role of Androgen Metabolism Genes CYP1B1, PSA/KLK3, and CYP11α in Prostate Cancer Risk and Aggressiveness

The causes of prostate cancer are not well known, but both genetic and environmental influences likely play a role. The prostate is an androgen-dependent organ and polymorphic variants in a number of genes involved in androgen metabolism have been implicated in prostate cancer risk. For example, we previously reported that SRD5A2, CYP3A4, and CYP3A5 variants may influence risk of developing prostate cancer or more aggressive disease (1, 2). Additional studies have suggested that variants in other genes involved in androgen metabolism or response, such as CYP1B1, prostate-specific antigen (PSA/KLK3), and CYP11α, may also play a role in prostate cancer risk.

CYP1B1 maps to chromosomal region 2p21-22 (3) and is involved in the activation of many procarcinogens and the hydroxylation of testosterone. Several polymorphisms in the CYP1B1 gene have been described (4), including two common single nucleotide polymorphisms (SNP): 355G/T (A119S) in exon 2, which is located in the substrate recognition site one (SR1S), and 4326C/G (L432V) in exon 3, which encodes the heme-binding domain. Functional studies suggest that these nonsynonymous SNPs may alter enzymatic activity and catalytic specificity of CYP1B1 (5). Several studies have evaluated the relationship between these CYP1B1 polymorphisms and risk of various cancers. These include colorectal (6), breast (7-11), ovarian (12), and prostate cancers (13-15). With regard to the latter, carrying the CYP1B1 355T allele, or the 4326G/G genotype, has been positively associated with prostate cancer among Japanese (14) and Caucasians (13), respectively. A more recent report, however, found an inverse association between a haplotype including 355T-4326C and sporadic prostate cancer (15).

The gene encoding PSA is regulated by the androgen receptor, which binds to androgen-responsive elements. Serum PSA level is used to screen for prostate cancer because it is often elevated in prostate cancer patients. PSA (also known as KLK3) is located on human chromosome 19q13.3-13.4, a region previously linked to advanced prostate cancer (16, 17). Both PSA/KLK3 −158A/G promoter variant alleles (i.e., −158A and −158G) have been associated with prostate cancer risk (18, 19). In addition, this variant has been positively associated with higher PSA levels in some studies, although other studies have found inverse or null associations (19-23). The −158A/G variant may be in linkage disequilibrium with the potential causal variants: a recent study of men without prostate cancer found that two other promoter polymorphisms, −4643G/A and −5412C/T (but not −158A/G), are associated with serum PSA levels and PSA promoter activity (22).

The CYP11α gene encodes the cholesterol P450 side-chain cleavage enzyme that catalyzes the conversion of cholesterol to pregnenolone mediating the first rate-limiting step in testosterone synthesis (24). CYP11α is located on chromosome 15q23-24 and the promoter region of this gene is believed to contain multiple cyclic AMP–regulated elements that are responsible for increasing basal transcriptional activity (25). Up-regulation of this enzyme may lead to an increase in androgen production (26). A microsatellite polymorphism (tttta)_n was identified in the promoter region of CYP11α, located 528 bp upstream from the initiation site of the translation of the gene (26). Recent studies have shown this polymorphism to be associated with increased risk of advanced prostate cancer (27), polycystic ovary syndrome (26, 28), and breast cancer (29).

In light of these existing results, here we aimed to further clarify the relation of polymorphisms in CYP1B1, PSA/KLK3, and CYP11α with prostate cancer and tumor aggressiveness using a sibling-based case-control population.

The study design and population have been described elsewhere (1). Briefly, siblings (n = 918; 439 cases and 479 controls from 413 families) were recruited from the major medical institutions in the greater Cleveland, Ohio area and from the Henry Ford Health System, Detroit, Michigan. Ninety-one percent of the subjects were Caucasian, 8% were African American, and 1% were Asian or Latino. Institutional Review Board approval was obtained from the participating institutions and all study participants gave informed consent.

Sibling sets consisted of men with prostate cancer diagnosed at age 73 or younger and at least one brother without prostate cancer. The case clinical characteristics at diagnosis (e.g., Gleason score, tumor stage) were obtained from medical records and their disease status was confirmed histologically. All controls were no more than 8 years younger than their brother's age at diagnosis. The disease status of unaffected brothers was further confirmed through testing of PSA levels whenever possible (93% of controls). If a control's PSA level was elevated (>4 ng/mL), they were referred for further evaluation by our collaborating urologists. We followed up these individuals to confirm that they remained disease-free. These criteria were selected in an attempt to minimize potential for misclassification of prostate cancer status amongst controls (30, 31). We chose to use this sibling-based study design to ensure that our control group was selected from the same genetic source population as the cases, and is thus not subject to population stratification from factors such as race (32).

Standard venipuncture was used to collect blood samples from all study participants in tubes with EDTA as an anticoagulant. Genomic DNA was extracted from buffy coats using the QIAmp DNA Blood kit (QIAGEN, Inc., Valencia, CA). All purified DNA samples were diluted to a constant DNA concentration of 5 ng/μL in 10 mmol/L Tris, 5 mmol/L EDTA buffer (pH 8).

CYP1B1. The CYP1B1_L432V polymorphism (+4326C/G; rs1056836) was detected by amplifying genomic DNA. PCR products were digested with restriction enzyme and the digested products were separated on an agarose gel. Analysis of CYP1B1_A119S (+355G/T; rs1056827) variant was done using the Amplifluor Uniprime SNP Detection System (Chemicon Intl., Serological Corp., Norcross, GA). The method included a “hot start” two-step PCR reaction followed by allelic discrimination protocol using the ABI Prism 7900HT (Applied Biosystems, Foster City, CA) for an end-point read and analysis.

PSA/KLK3. The PSA/KLK3 G−158A (rs266882), A−4643G (rs925013), and T−5412C (rs2739448) polymorphisms were detected by amplifying genomic DNA. PCR products were digested with restriction enzyme and the digested products were separated on an agarose gel.

CYP11α. The CYP11a (tttta)_n microsatellite polymorphism was detected by amplifying genomic DNA. The forward primer was fluorescently labeled by Beckman color green (D3). PCR product fragment lengths were analyzed on a CEQ 8000 genetic analysis system (Beckman Coulter, Fullerton, CA). The number of (tttta) repeats was confirmed by sequencing at least one sample of each PCR product length that we observed.

To ensure quality control of genotyping results, 5% of samples were randomly selected and genotyped by a second investigator and 1% was sequenced using a 377 ABI automated sequencer (all results were concordant).

A χ² goodness of fit test was used to assess deviation from Hardy-Weinberg equilibrium for genotype frequencies of each variant. For the three SNPs in PSA/KLK3 and two SNPs in CYP1B1, linkage disequilibrium was calculated using HAPLOVIEW (http://www.broad.mit.edu/personal/jcbarret/haploview/index.php) in Caucasian or African American controls (randomly selecting one control from each family). Next, we reconstructed diplotypes (i.e., joint haplotypes) for both PSA and CYP1B1 with the program PHASE (33), initially ignoring relations among siblings. We then estimated diplotypes incorporating relations for families with >2 siblings in the study, using MERLIN (MERLIN v1.0-α, http://www.sph.umich.edu/csg/abecasis/Merlin). The assigned diplotypes were identical using these two approaches.

Conditional logistic regression (using family as the matching variable, and a robust variance estimator that incorporates familial correlations) was used to estimate odds ratios (OR) and corresponding 95% confidence intervals (95% CI) for the relation between genotypes/diplotypes and prostate cancer. The rare diplotypes (observed in <4% of the entire sample) were collapsed into one group. We also did analyses by grouping diplotypes according to the number of haplotype of interest. The regression model was adjusted for potential confounding by age. To investigate the potential effect of genotypes/diplotypes on disease aggressiveness, we stratified the analyses by the cases clinical characteristics at diagnosis (n = 860). A total of 58 individuals (cases and controls) did not have a complete clinical data and, therefore, were not included in the analyses stratified by disease aggressiveness. Following Rebbeck et al. (34), aggressiveness was defined as “low” if the Gleason score of a case was <7 and the tumor category was <T_2C, and “high” if the Gleason score was ≥7 or the tumor category was ≥T_2C. The tumor category reflects the tumor-node-metastasis system. All P values are from two-sided tests, and all analyses were undertaken with SAS software (version 8.2, SAS Institute, Cary, NC).

The present study evaluated the potential association between prostate cancer and polymorphisms in three genes involved in androgen metabolism: CYP1B1, PSA/KLK3, and CYP11α. The mean ages of cases and controls were 61 and 63 years, respectively, as expected based on our recruitment of slightly older controls. Also as expected, cases had much higher PSA levels in comparison with the controls (mean PSA: 14.8 versus 2.2). About half of our cases were diagnosed with more advanced prostate cancer: 191 (45%) had Gleason scores ≥7, and 55 (13%) have tumor stage ≥T_2C (Supplementary Table S1). The variants in PSA and CYP1B1 were in Hardy-Weinberg equilibrium and had strong linkage disequilibrium within genes (D′ > 0.8).

Table 1 provides ORs for the association between variants in CYP1B1 and prostate cancer. Among all study subjects, there was no association between the two variants (G355T and C4326G) and prostate cancer, although there was a weak positive association for the T-C/T-C diplotypes (OR, 2.17; 95% CI, 1.01-4.67; P = 0.05). This result was strengthened among men with more aggressive disease: in comparison with men homozygous for the G-G haplotype (i.e., for SNPs G355 and 4326G), men carrying the T-C haplotype had statistically significantly increased risk, except for the diplotype G-C/T-C (Table 2). When we grouped diplotypes according to the number of T-C haplotypes, carrying one and two copies of this haplotype was associated with aggressive prostate cancer, with ORs of 1.99 (95% CI, 1.15-3.47; P = 0.01) and 4.00 (95% CI, 1.36-11.8; P = 0.01). This effect seemed to be primarily due to the G355T SNP. Specifically, in comparison with the GG genotype, carrying one or two T alleles gave ORs for prostate cancer of 1.90 (95% CI, 1.09-3.31; P = 0.02) and 3.73 (95% CI, 1.30-10.0; P = 0.009), respectively (P for trend = 0.004). This finding is in agreement with studies of prostate (14) and breast and lung cancers (35). Other studies, however, reported an inverse association between prostate cancer and a haplotype including 355T-4326C (15), and a positive association with the C4326G SNP alone (13). These conflicting results might be explained by the very small sample size included in the latter study (n = 50) and our observation that not all haplotypes carrying the 355T allele led to an increased risk. Whereas functional studies support the direct role of these variants on disease risk (5), there may be a causal variant residing on or near the corresponding haplotype. No noteworthy associations were observed among men with low aggressive prostate cancer (Table 2).

When looking at KLK3/PSA, no associations were observed for the three SNPs (A−158G, A−4653G, and T−5412C), or their haplotypes, among the entire sample (Table 3). However, we observed a statistically significant association between the A−158G variant and prostate cancer among men with nonaggressive disease. In comparison with men homozygous for the −158A allele, carrying the GG genotype gave an OR of 2.71 (95% CI, 1.06-6.94; P = 0.04). And there was a slight trend across genotypes, as each additional copy of the 158G variant seemed to increase the respective ORs (P for trend = 0.03). Similar, albeit weaker, results were observed for the PSA haplotypes: men with diplotypes carrying two copies of the −158G variant exhibited increased risk of disease (though not statistically significant; Table 2). In contrast, within the higher aggressiveness stratum, all nonreferent diplotypes (i.e., not AAT/AAT) were weakly inversely associated with disease (P ≥ 0.04). Analyses for the number of GGC haplotypes did not materially change the above results (Table 3). This is in general agreement with two previous studies (18, 36); however, another study observed a positive association between carrying two copies of the −158A variant and disease (19). Of course, the −158A/G SNP may be in linkage disequilibrium with the causal variants in the PSA/KLK3 promoter region. However, we found no associations for two other promoter polymorphisms (G−4643G and C−5412T). In addition, three-locus (A−158G, G−4643A, and C−5412T) haplotype analyses did not improve the results observed for the single A−158G variant. To this end, functional studies suggest that the proliferation, migration, and metastasis of prostate cancer cells may be due to the enzymatic activity of PSA/KLK3 (37). Our findings suggest that the −158G/−158G may be involved in disease risk via this enzymatic pathway reflecting the effect of other functional variants.

Finally, we observed no associations between the CYP11α (tttta) repeat and prostate cancer risk or tumor aggressiveness (Table 3). Here we recoded genotypes for CYP11α as three groups: noncarriers of 4 (tttta) repeat allele (none), subjects carrying one 4 (tttta) repeat allele (one copy), and homozygotes for 4 (tttta) repeat alleles (two copies). Using other codings for the genotypes did not alter our null results (not shown). Our observation of no overall association between prostate cancer and the CYP11A (tttta)_n variant is consistent with a previous study (27). Nevertheless, Kumazawa et al. (27) did observe that the absence of the (tttta)_n variant was associated with an increased risk of stage D disease (i.e., T_1-4N₁M_0-1 or T_1-4N_0-1M₁ according to the Whitmore-Jewett system; ref. 38). Our inability to replicate their results may in part reflect the small number of men in our study with such high-stage prostate cancer.

By using a family-based design, the current study is not susceptible to confounding bias due to population stratification (39). Nevertheless, studying brothers with and without prostate cancer can lead to overmatching and subsequent reduced power for detecting genetic effects (32).

In summary, we found evidence that polymorphisms in CYP1 and PSA/KLK3 may affect the risk of prostate cancer among men with more and less advanced disease, respectively. Moreover, these genes may contain other functional variants that affect the aggressiveness of prostate cancer.