Novel associations of UDP-glucuronosyltransferase 2B gene variants with prostate cancer risk in a multiethnic study

Background We have previously shown that a functional polymorphism of the UGT2B15 gene (rs1902023) was associated with increased risk of prostate cancer (PC). Novel functional polymorphisms of the UGT2B17 and UGT2B15 genes have been recently characterized by in vitro assays but have not been evaluated in epidemiologic studies. Methods Fifteen functional SNPs of the UGT2B17 and UGT2B15 genes, including cis-acting UGT2B gene SNPs, were genotyped in African American and Caucasian men (233 PC cases and 342 controls). Regression models were used to analyze the association between SNPs and PC risk. Results After adjusting for race, age and BMI, we found that six UGT2B15 SNPs (rs4148269, rs3100, rs9994887, rs13112099, rs7686914 and rs7696472) were associated with an increased risk of PC in log-additive models (p < 0.05). A SNP cis-acting on UGT2B17 and UGT2B15 expression (rs17147338) was also associated with increased risk of prostate cancer (OR = 1.65, 95% CI = 1.00-2.70); while a stronger association among men with high Gleason sum was observed for SNPs rs4148269 and rs3100. Conclusions Although small sample size limits inference, we report novel associations between UGT2B15 and UGT2B17 variants and PC risk. These associations with PC risk in men with high Gleason sum, more frequently found in African American men, support the relevance of genetic differences in the androgen metabolism pathway, which could explain, in part, the high incidence of PC among African American men. Larger studies are required.

Methods: Fifteen functional SNPs of the UGT2B17 and UGT2B15 genes, including cis-acting UGT2B gene SNPs, were genotyped in African American and Caucasian men (233 PC cases and 342 controls). Regression models were used to analyze the association between SNPs and PC risk.
Results: After adjusting for race, age and BMI, we found that six UGT2B15 SNPs (rs4148269, rs3100, rs9994887, rs13112099, rs7686914 and rs7696472) were associated with an increased risk of PC in log-additive models (p < 0.05). A SNP cis-acting on UGT2B17 and UGT2B15 expression (rs17147338) was also associated with increased risk of prostate cancer (OR = 1.65, 95% CI = 1.00-2.70); while a stronger association among men with high Gleason sum was observed for SNPs rs4148269 and rs3100.
Conclusions: Although small sample size limits inference, we report novel associations between UGT2B15 and UGT2B17 variants and PC risk. These associations with PC risk in men with high Gleason sum, more frequently found in African American men, support the relevance of genetic differences in the androgen metabolism pathway, which could explain, in part, the high incidence of PC among African American men. Larger studies are required.

Background
Prostate cancer is the second leading cause of cancerrelated deaths in men, after lung cancer [1]. The incidence of prostate cancer has increased over the past twenty years and African American men have been disproportionally affected compared to other racial/ethnic groups [2][3][4][5][6][7]. In the U.S., the incidence of prostate cancer among African Americans is more than 60% higher than in Caucasians, and the mortality rate in African Americans is twice that of Caucasian men [8,9]. Although differences in incidence and mortality rates may be due, in part, to race/ethnicity, socioeconomic conditions and availability of health care [10], familial aggregation studies suggest that genetic factors may also be contributing to prostate cancer demographic disparity. Candidate gene approaches involving hormone metabolic pathways have been examined in prostate cancer association studies, however results from these studies have not been replicated [11,12]. Nonetheless, current therapies are primarily targeted at specific androgen biosynthetic pathways [13], thus, improved knowledge on genetic variants associated with both androgen metabolism and prostate cancer risk is important.
The UDP-glucuronosyltransferase (UGT) genes code for enzymes that convert a diverse group of xenobiotic and endobiotic substances into lipophilic compounds, facilitating clearance from the body as part of the phase II liver detoxification system [14]. These enzymes are subdivided into families according to similarities in amino acid sequence and target substrates dictated by the specificity of their amino-terminal ends [15]. A specific subfamily, UGT2B, includes two enzymes, UGT2B17 and UGT2B15, which are also expressed extrahepatically in the basal and luminal epithelium of the prostate, respectively. These enzymes exhibit specificity for androgen metabolites such as testosterone, dihydrotestosterone (DHT), androsterone (ADT), and androstane-3α,17β-diol (3β-diol) in prostate tissue and cell lines [16,17].
Functional polymorphisms of the UGT2B gene family such as the copy number variant (CNV) of UGT2B17, and the aspartic acid (D or nucleotide G) to tyrosine substitution (Y or nucleotide T) found in codon 85 of the UGT2B15 gene (UGT2B15 D85Y , rs1902023) have been identified [18,19]. Studies suggest that men with 0 copies of the UGT2B17 gene are unable to break down testosterone through the UGT2B pathway and subsequently secrete negligible amounts of urinary testosterone compared to men with at least one copy of UGT2B17 [18,20]. Experimental evidence showed that the minor allele (T) of the UGT2B15 variant, UGT2B15 D85Y , causes the enzyme to have an increased V max activity when compared to the presence of the major allele (G) [19]. Subsequently, the resulting phenotype of this UGT2B15 polymorphism is a quicker androgen metabolite clearance which may raise the "effective" amount of steroids within the prostate and decrease risk for prostate cancer [21]. These two major UGT2B variants (the CNV in UGT2B17 and the polymorphism in UGT2B15) have been evaluated in relation to prostate cancer risk, with inconsistent findings [22][23][24][25][26][27][28][29][30][31][32]. Discrepancies could have been due to the genetic heterogeneity of the populations studied, as well as variable sample sizes of these populations. We have recently shown that individuals with a major allele (G) of the UGT2B15 D85Y polymorphism (rs1902023) have higher risk of prostate cancer when compared to individuals homozygous for the "rapid clearance" minor allele (Y) [33]. In the same study, the UGT2B17 CNV showed no association with prostate cancer risk [33]. Recently, an additional 7 novel UGT2B15 SNPs that are in strong linkage disequilibrium (LD) with the UGT2B15 D85Y gene variant, have been identified by re-sequencing the promoter and exon one regions of the UGT2B15 gene, using DNA samples from Yoruba (YRI), CEPH/European (CEU), and Japanese/Chinese (ASN) populations [34]. Most of these UGT2B variants have not been evaluated in relation to prostate cancer risk in population-based studies or in studies that included African American men. In this present work, we examined associations between functional SNPs of UGT2B17, UGT2B15 and three other related UGT2B SNPs, and prostate cancer risk among African American and Caucasian men.

Study population
The details of participant accrual for this case control study have been previously reported [35]. In brief, male subjects from the Durham Veterans Affairs Medical Center (DVAMC) in Durham, North Carolina, who were undergoing a prostate needle biopsy between January 2007 and October 2011, were consecutively contacted in a hospital-based, case control study. Eligibility criteria for cases included age >18 years, undergoing a prostate biopsy for concerns of potential prostate cancer after presentation with elevated PSA and/or abnormal digital rectal examination, and prostate cancer positive classification by pathological review of biopsy tissue. Of the 759 men with a biopsy indication who were screened for eligibility, 539 (759/539 = 71% participation rate) provided written consent to participate. Twenty two men elected not to follow through and of the 517 men that underwent a biopsy, 233 had a biopsy with histological evidence of prostate cancer. Controls were recruited from the DVAMC Internal Medicine Clinic; eligibility criteria were age >18 and having a PSA test conducted at the DVAMC within the same time frame, but not recommended to undergo biopsy. Of the 768 men who met eligibility criteria for controls, 377 provided written consent (768/377 = 49% participation rate). Questionnaires were administered to prostate cancer cases prior to biopsy and to controls to assess risk factors including race and age. Institutional Review Board approval was obtained at North Carolina Central University, Duke University, and the DVAMC and all patients signed an informed consent prior to enrollment.
DNA was isolated from peripheral blood by standard DNA isolation (Qiagen Inc., Valencia, CA, U.S.A.) and quantified by ultraviolet spectrophotometry. Prior to genotyping, DNA concentration was determined using PicoGreen assay (Life Technologies, Gaithersburg, MD) and measured using the fluorescence intensity measurements plotted against a standard curve that was generated from the average fluorescence intensity of standards run

Statistical analysis
We examined whether genetic variants of UGT2B17 and UGT2B15 genes were associated with prostate cancer risk and whether these associations varied by race and prostate cancer grade. Low grade prostate cancer was defined as Gleason sum <8 and high-grade prostate cancer was defined as Gleason ≥ 8. Potential confounders (age, height, and BMI) were normally distributed overall and were treated as continuous variables in modeling. Race was self-reported and categorized as African American, Caucasian and other (Native American and Latino). Family history of prostate cancer included maternal lineage (maternal uncle) or paternal lineage (father, brother or paternal uncle) (yes or no) and these combined. Data on tobacco use at the time of enrollment was also self-reported as yes or no. Descriptive statistics (means, SD), and percentages for cases and controls were estimated using X 2 tests for categorical variables, and Wilcoxon rank-sum test for continuous variables. Unconditional logistic regression was used to estimate the odd ratios (ORs) and 95% confidence intervals (95% CI) for the association between genotypes and prostate cancer risk. Multinomial logistic regression models were used to explore whether associations between genotypes and prostate cancer risk varied by grade of prostate cancer at diagnosis, using controls as reference. Confounders adjusted for in all models were age, race, and BMI. All analyses were done using SAS version 9.3 (SAS Institute, Inc., Cary, NC). Table 2 describes the clinical characteristics of the study participants. Prostate cancer cases (n = 233) and controls (n = 342) were comparable in height (p = 0.31), however controls (mean BMI = 30.77, SD = 6.05) had slightly higher BMI than cases (mean BMI = 29.25, SD = 5.60, p = 0.002) and were slighter younger than cases (p = 0.005). African American men comprised 63% of the cases and 41% of the controls (p <0.0001). A family history of prostate cancer was reported by 14% of cases and 10% of the controls. (p = 0.16). Most cases and controls reported tobacco use (p = 0.43). Ninety-one percent of the cases had low-grade prostate cancer (Gleason sum < 8) vs. 9% of cases with high grade prostate cancer (Gleason sum ≥ 8).
To investigate the extent to which these SNPs were associated in our subjects we calculated correlation coefficients among the controls as shown in Table 6. The r 2 values for the UGT2B15 variants rs1580083, rs1960773, rs13112099, rs7686914, rs7696472 and rs9994887 ranged from 0.62 to 0.99/1.00 (p < 0.0001). UGT2B15 variants rs4148269 and rs3100 were also strongly correlated (r 2 = 0.85, p < 0.0001). Including both SNPs into the same regression model to evaluate whether one SNP would have a stronger association with PC risk did not yield additional insights. Notably, these two SNPs showed no correlation with the other SNPs that were highly associated with each other. UGT2B17 SNP rs7435827 was moderately correlated with UGT2B15 SNP rs2168047 (r 2 = 0.36, p < 0.0001) and UGT2B17 SNP rs7671342 (r 2 = 0.32, p < 0.0001), and strongly correlated with UGT2B17 SNP rs59678213 (r 2 = 0.83, p < 0.0001); however these SNPs showed no association with prostate cancer risk. UGT2B17 SNP rs59678213 was inversely associated with UGT2B17 SNPs rs2168047 (r 2 = −0.47, p < 0.0001), rs7686008 (r 2 = −0.29, p < 0.0001) and rs7671342 (r 2 = −0.49, p < 0.0001). No correlations were observed for the cis-acting UGT2B SNPs that were identified through the recent UGT2B SNPs discovery study [38]. All SNPs that showed associations with prostate cancer risk were in Hardy-Weinberg equilibrium in control samples, except for UGT2B15 SNP rs3100.

Discussion and conclusion
In this study we examined associations between functional SNPs of the UDP-glucuronosyltransferase 2B (UGT2B) genes and prostate cancer risk. After adjusting for age and BMI, we found that two UGT2B15 SNPs, rs4148269 and rs3100, were associated with a two-fold increase risk of prostate cancer and the association was more apparent in African American men. In addition, SNPs rs9994887, rs13112099, rs7686914 and rs7696472 were also associated with increased risk of prostate cancer. These associations persisted either in low or high-grade prostate cancer lesions for SNPs rs9994887, rs13112099, rs7686914, and rs7696472. However, SNPs rs4148269 and rs3100 were only associated with increased risk of prostate cancer in high-grade prostate cancer lesions. These findings are novel and support the hypothesis that these SNPs may affect UDP-glucuronosyltransferase enzyme function, presumably by increasing the efficiency of androgen metabolite clearance compared to enzyme products of the UGT2B wild type genotype, as has been shown for UGT2B15 SNP rs1902023 [21].
Associations between functional SNPs of the UGT2B15 and UGT2B17 genes and prostate cancer risk have been previously examined and have rendered conflicting results [22][23][24][25][26][27][28][29][30][31][32]. Discrepancies could be due in part to racial/ethnic differences in the populations studied but also to sample size and allele dosage. Comparison of two studies with different sample sizes revealed contradictory results: no associations between SNP UGT2B15 D85Y and prostate cancer risk were found in a study with a large sample size [9], whereas a study with a smaller sample size found an increased risk of prostate cancer associated with the    [32]. In reference to allele dosage, we previously concluded [33], concordant with previous reports [31,32,39], that homozygous carriers of the dominant UGT2B15 D85Y (G) allele missense polymorphism, (rs1902023), had an almost three-fold higher risk of prostate cancer. Our current results further confirm that associations with UGT2B15 polymorphisms are allele dose-dependent and may also be dependent on race/ethnicity. Furthermore, our study includes a large and multiethnic sample size in comparison to previous reports. Some of our stratified analyses revealed that UGT2B15 wild type alleles were significantly associated with increased risk of prostate cancer in African American men, but not in Caucasian men, suggesting that the wild type genotype may confer protection against prostate cancer in Caucasian men. While these findings require replication in larger studies, we provide new evidence that wild type genotypes for the UGT2B15 gene more frequently found in African American men may increase the risk of prostate cancer, and thus may contribute to the present racial/ethnical disparities seen in prostate cancer incidence. Few studies have assessed the association of UGT2B genes with prostate cancer risk in African Americans. Park et al. [25] were one of the first to report that associations between UGT2B17 polymorphisms and increased risk of prostate cancer were linked to race/ethnicity; they showed that Caucasian men carrying a UGT2B17 deletion polymorphism had a two-fold increased risk of prostate cancer, but not Black men who carry the same deletion, compared to noncarriers [25]. Subsequent work supported these findings [22,26], which were further confirmed by others [23,24]. Our results show no significant associations of UGT2B17 polymorphisms with prostate cancer risk in neither Caucasian nor African American men. Our findings that UGT2B17 SNP rs59678213 was not associated with prostate cancer risk are consistent with those of others [22,24].
Four of the UGT2B15 SNPs that conferred prostate cancer risk in this study are among seven SNPs (rs1580083, rs1960773, rs9994887, rs13112099, rs7686914, rs7696472, and rs1120265 [not in study]) that are in high linkage disequilibrium (LD) with UGT2B15 D85Y SNP rs1902023 [34]. The phenotype of the variations, established with reporter gene assays in the liver cancer cell line HepG2, showed that transcriptional activity was higher for constructs containing the major alleles when compared to constructs containing the minor alleles. When a construct with the minor allele of rs9994887 was transfected into a prostate cancer cell line, LNCaP, transcriptional activity was decreased by approximately 14%, compared to constructs containing the major allele [34], which suggested that rs9994887 may influence gene expression in the prostate [34]. Further experiments would have to be done to determine the impact of these variants on the kinetics and efficacy of UGT2B enzymes. Interestingly, the other three SNPs that showed associations with increased prostate cancer risk, rs13112099, rs7686914, and rs7696472 were highly correlated with rs9994887 (r 2 = 0.99, 0.99, and 0.97, respectively). In addition, the three SNPs are all located upstream of rs9994887 (−1139 bp relative to translation start) while the other two SNPs are located downstream. While the impact of these variants on UGT2B15 enzyme kinetics has not been reported, the associations suggest that the phenotype of the enzyme may be similar to that of the UGT2B15 85Y variant or rs1902023 [19]. SNPs rs4148269 and rs3100 also demonstrated strong associations with decreased risk of PC, but did not show any correlation with the promoter SNPs in this study population. SNP rs4148269 has been characterized as a T523K amino acid substitution [40], however a comprehensive analysis of the enzyme phenotype has not been reported. This suggests that rs3100, which is highly correlated (r 2 = 0.85) with rs4148269, may be important in the etiology of PC. Results from reporter gene assays showed that the major allele of rs3100 is associated with significantly higher gene expression in LNCaP cells when compared to constructs made with the minor allele [36].
The lack of association observed for the UGT2B17 SNPs and PC risk was unexpected, as previous reports suggested that the UGT2B17 enzyme plays a major role in testosterone glucuronidation [20]. However results from a recent  report comparing UGT2B17 and UGT2B15 mRNA and protein expression, suggest that UGT2B15 may be negatively regulated in naïve and castration-resistant tumors while undetectable in lymph node metastases [41]. More in vitro analyses are required to understand the functional impact of these genetic variations. Nonetheless this is the first time these SNPs are reported in relation to prostate cancer risk.
The main limitation of this study is the sample size which limited our ability to simultaneously examine race/ethnicity in the associations with SNPs and prostate cancer grade. The low numbers of Caucasians may have limited our ability to detect significant associations within this group given that they were less likely to present a high grade PC tumor. The small sample size could also have been a contributing factor in the lack of significance for the cross-product term for the SNPs and race, however race/ethnicity was self-reported and we did not measure ancestral markers, which could indicate the percentage of African American ancestry of each subject. Also, we cannot exclude the possibility of potential confounding due to differences in MAFs between races. Nonetheless our study was conducted at the DVAMC which provides medical care for all former military personnel. This type of setting has important advantages. The Veterans Affairs Medical Centers (VAMC) are considered an equal-access health care settings that may control for health care availability among various U.S. ethnicities [42]. Moreover, the VAMCs provide a setting for PC case-control studies as this is the leading cancer reported in Veterans Affairs Central Registry (VACCR) [43].
Despite these limitations, in the current case-control study of African American and Caucasian men, we observed elevated risk of prostate cancer in African American men who carry the wild type of the functional UGT2B15 gene variants. Future studies with larger sample sizes are needed to confirm these findings.