The PXR rs7643645 Polymorphism Is Associated with the Risk of Higher Prostate-Specific Antigen Levels in Prostate Cancer Patients

Levels of enzymes that determine testosterone catabolism such as CYP3A4 have been associated with prostate cancer (PCa) risk. Although some studies have related CYP3A4*1B allele, a gene polymorphism that modifies CYP3A4 expression level, with PCa risk, others have failed, suggesting that additional genetic variants may be involved. Expression of CYP3A4 is largely due to the activation of Pregnane X Receptor (PXR). Particularly, rs2472677 and rs7643645 PXR polymorphisms modify CYP3A4 expression levels. To evaluate whether PXR-HNF3β/T (rs2472677), PXR-HNF4/G (rs7643645), and CYP3A4*1B (rs2740574) polymorphisms are associated with PCa a case control-study was performed. The multiple testing analysis showed that the PXR-HNF4/G polymorphism was associated with higher levels of prostate-specific antigen (PSA) in patients with PCa (OR = 3.99, p = 0.03). This association was stronger in patients diagnosed at the age of 65 years or older (OR = 10.8, p = 0.006). Although the CYP3A4*1B/*1B genotype was overrepresented in PCa patients, no differences were observed in the frequency of this and PXR-HNF3β/T alleles between controls and cases. Moreover, no significant association was found between these polymorphisms and PSA, Gleason grade, or tumor lymph node metastasis.


Introduction
Prostate cancer (PCa) is the second most common cancer worldwide in males and one of the most common causes of death in men (IARC, 2008). The etiology of the disease involves several factors, including ethnicity, older age, family history, and genetic and environmental factors [1,2]. It has been established that steroid hormone levels, in particular androgens, affect the risk of developing PCa [3,4]. Testosterone and predominantly its metabolite dihydrotestosterone interact with the androgen receptor, which leads to the expression of genes involved in the growth of the prostate and the proliferation of prostate cancer cells [5]. Several cytochromes P450s (CYPs) are involved in the synthesis and catabolism of hormones, such as testosterone [6]. The bioavailability of dihydrotestosterone is decreased by CYP3A4, which mediates the 2b-, 6b-, and 15b-hydroxylation of testosterone in the liver and prostate [7][8][9]. Therefore, it has been hypothesized that low levels and/or decreased CYP3A4 activity might result in a lower capacity to inactivate testosterone favoring its conversion to dihydrotestosterone and increasing the risk of developing PCa. In fact, decreased expression of CYP3A4 has been found in prostatic tissues from PCa patients compared to 93% for benign epithelium, and only 75% of prostate tumors expressed CYP3A4 [10].
Interindividual variation in CYP3A4 levels of up to 60-fold have been reported [11,12], and it has been suggested that most of the variability can be explained by genetic factors [13]. The CYP3A4 gene is highly polymorphic, and to date, 43 different CYP3A4 polymorphisms have been reported (http://www.cypalleles.ki.se/), of which CYP3A4*1B is one of the most common. CYP3A4*1B is a single nucleotide polymorphism (SNP) (rs2740574) that introduces an A to G substitution at position 2290, which is located in the nifenipine specific response element of the promoter of the CYP3A4 gene. While some studies have reported associations between the CYP3A4*1B allele and higher clinical grade of PCa [14,15], others have failed to observe an association between the presence of the CYP3A4*1B polymorphism and prostate cancer susceptibility [16,17]. Controversial data regarding the association of CYP3A4*1B and CYP3A4 activity have also been reported [18,19], suggesting that other genetic variants may be involved, such as transcription factors that mediate CYP3A4 expression. Pregnane X receptor (PXR, NR1I2) is a member of the steroid nuclear receptor family of ligand activated transcription factors. Activated-PXR forms a heterodimer with 9-cis-retinoic acid X receptor (RXR) and binds to a nuclear receptor response element in the 59-flanking region of its target genes. Induction of CYP3A4 is largely due to the activation of PXR [20]. Therefore, genetic variants of PXR that alter PXR protein levels or its transactivation potential may have an important impact on CYP3A4 expression. Several PXR polymorphisms have been described to date, but only a few have an effect on CYP3A4 function. Among them, we found rs2472677 and rs7643645. The rs7643645 SNP is located in the HNF4 binding site of the promoter of the PXR gene and has been associated with decreased PXR and CYP3A4 mRNA levels as well as CYP3A4 activity. The rs2472677 variant is located in the HNF3b binding site of the same promoter and results in increased PXR mRNA levels as well as basal CYP3A4 activity [21]. For practical purposes, in the present study the rs7643645 variant will be called PXR-HNF4/G or PXR-HNF4/WT and the rs2472677 variant will be called PXR-HNF3b/T or PXR-HNF3b/WT.
So far, there have been no reports in the literature exploring the possible association between PXR polymorphisms and PCa. Therefore, we performed a case-control study to investigate whether CYP3A4*1B, PXR-HNF4/G, and PXR-HNF3b/T allele variants are associated with a risk of developing PCa in Mexican men.

Study population
The present study used a hospital-based case-control design. The case group was recruited from the National Institute of Cancer and included 99 patients with a histologically confirmed diagnosis of PCa. Clinical characteristics, such as Gleason grade, prostate-specific antigen (PSA) levels at the time of diagnosis, digital rectal examination (DRE; according to the American Urological Association recommendations), tumor lymph node metastasis (TNM) [22], and age at the time of diagnosis were obtained from medical records. Categories of clinical characteristics were defined as follows: PSA two groups, (cut point, 10 ng/ mL); Gleason grade two groups (cut point, 7); and TNM two groups (TNM#2 and TNM$3) [15]. The control group consisted of 144 patients with no history of any cancer, including prostate cancer, with a PSA,4.0 ng/mL, normal DRE, and was recruited from the Juárez Hospital. All subjects were unrelated men (selfreporting) between 60 and 76 years of age. The National Institute of Cancer and the Juárez Hospital Ethics Committees approved the present study and written informed consent was obtained from all subjects. The complete clinical and genetic per-patient data can be found in the supporting information (Table S1 and S2).

DNA extraction
Genomic DNA was isolated from 5 ml of whole blood using a sodium perchlorate/chloroform extraction method. Briefly, DNA was prepared by combining each blood sample with 35 mL of lysis

Genotyping
Genotyping of CYP3A4*1B, PXR-HNF3b, and PXR-HNF4 was conducted by real-time PCR using a StepOne Real-Time PCR System with TaqMan Universal PCR Master Mix (Applied Biosystems, USA). PCR was conducted using 5 ml of TaqMan Universal PCR Master Mix, 0.25 ml of primer-probe mix (containing 36 mM of each primer and 8 mM of dye-labeled probe), and 20 mg of DNA template. The final reaction volume was brought to 10 ml with H 2 O. Forty PCR cycles of the following parameters were used: initial denaturalization at 95uC for 10 min, 15 s at 92uC, and then 60uC for 1 min. After each amplification an allelic discrimination was made to determine the genotype of each subject. The primers and probes sequences used for CYP3A4 were as follows:

Statistical analysis
Statistical analysis was conducted using the STATA statistical package (version 10.1, STATA Corp., College Station, TX). Ttest, X 2 test or Fisher's exact test were used to evaluate whether the distribution of genotype frequencies of CYP3A4*1B, PXR-HNF4, and PXR-HNF3b varied among cases and controls. For the comparison of clinical characteristics in the case group, ORs were calculated as an estimate of relative risk and 95% confidence intervals (CIs) were calculated using a bivariate logistic model. A value of p,0.03 was considered statistically significant after multiple testing adjustment by using the Bonferroni correction. The X 2 test was also used to assess deviations of allelic frequencies from Hardy-Weinberg equilibrium. The interaction between alleles was analyzed using the software Plink V 1.07. The number (n) of the case population for each association with clinical characteristics is indicated in the tables.
The sensitivity and specificity from significant models were estimated [23]. A cutoff of 50% for the classification of the event and the Receiver Operating Characteristic (ROC curve) were used. All calculations were performed using STATA as post logistic model estimation.

Results
Clinical characteristics of the studied subjects were obtained from medical records and are presented in Table 1. No differences were observed in marital status or age between the two groups. As expected, prostate cancer patients presented significantly higher levels of PSA than in the control group (167 vs. 1.73 ng/mL, respectively) as well as a higher score for DRE grade III.

Analysis between case and control groups
The genotype frequencies of CYP3A4 and PXR between prostate cancer patients and controls are shown in Table 2. Alleles for PXR were in Hardy-Weinberg equilibrium (HWE), but CYP3A4*1B was only in HWE in the control group (data not shown). The CYP3A4*1B/*1B genotype was only present in prostate cancer patients. No differences were observed for CYP3A4*1A/*1B or CYP3A4*1A/*1A genotypes between both groups. Moreover, when PXR polymorphisms were compared between case and control groups, no differences in genotype frequencies for either PXR-HNF3b or PXR-HNF4 alleles were observed.
We then determined the distribution of the combined CYP3A4 and PXR allelic variants. Individuals with the combined genotypes CYP3A4*1B/*1B and PXR-HNF4G/G, or CYP3A4*1B/*1B and Table 3. Distribution of combined CYP3A4 and PXR allelic variants.   PXR-HNF3bWT/WT, were only present in the case group. For the other combinations, no differences were observed between control and case groups (Table 3).

Analysis among cases
Stratification of the CYP3A4 and PXR genotypes and PSA, Gleason grade, and TNM score among prostate cancer patients are shown in Figure 1 and Table S3. There were no differences in the CYP3A4 genotype by comparison of PSA or Gleason grade. Twenty five percent of the prostate cancer patients with a TNM$ 3 were heterozygous for the CYP3A4 variant and had an OR of 3.16 compared to the *1A/*1A genotype, but this difference was not statistically significant (p = 0.10) (Table S3). For the PXR-HNF4 genotype, an association of PXR-HNF4/G variant with higher PSA levels was observed in case group (Figures 1 and 2). Individuals with PSA.10 ng/ml were WT/G (49.2%) and G/G (34.92%) with ORs of 2.46 (p = 0.10) and 3.99 (p = 0.03), respectively. On the other hand, sixty two percent of the cases with a TNM$3 were heterozygous for the PXR-HNF4 allele with an OR of 2.96, but this result was not statistically significant. No associations were observed for PXR-HNF4 genotypes with the Gleason grade. The PSA, Gleason grade, and TNM score values were not associated with the PXR-HNF3b genotype among PCa patients. In order to determine the power of the present study for PXR-HNF4 polymorphism a power analysis was performed, and a 1-b value of 84 was obtained (Figure 3).
A stronger association between the CYP3A4 genotype and Gleason and TNM grades in patients diagnosed at the age of 65 years or older has been reported previously [15]. Therefore, we investigated the association of CYP3A4 and PXR-HNF4 genotypes with clinical characteristics among individuals diagnosed at the age of 65 years or older. In order to compare the results with those previously reported, a dominant model was used (i.e. WT versus heterozygous and homozygous for the variant).
No association between CYP3A4 genotypes and PSA or Gleason grade were observed. However, men with *1B allele showed an increased risk of having a higher TNM score (OR = 2.2), but the statistical analysis indicated that this association was not significant (Table 4). Table 5 shows ORs comparing PXR-HNF4-WT to PXR-HNF4 WT/G and PXR-HNF4 -G/G in combination (dominant model). The data indicate that men with the G variant showed an increased risk of having higher PSA levels (OR = 10.8; p = 0.006). Sensitivity and specificity analysis indicate that 84% of individuals over 65 years old with WT/G or G/G genotype will have a high risk to present higher levels of PSA.
a Comparing those with WT versus those with G/G and WT/G genotypes (dominant model). * Logistic model. n, No. of subjects.
No associations with Gleason grade and TNM score were identified. Taken together, these results clearly indicate that PXR-HNF4/G polymorphisms increase the risk of having higher PSA levels among cancer prostate patients.

Discussion
The etiology of prostate cancer still remains unclear and involves several factors, including genetic determinants. Therefore, the identification of genetic risk factors for prostate cancer susceptibility is important. Many studies have focused on CYP3A4 gene polymorphisms because this enzyme participates in testosterone metabolism. In the present case-control study, we investigated whether CYP3A4*1B variant was associated with several clinic characteristics of prostate cancer. The CYP3A4*1B variant was not in HWE in the case group most likely due to the excess of homozygous CYP3A4*1B genotype presented in this group. Deviation from HWE should not be unexpected, particularly when an allele is associated with a disease, which is the case in this study. In fact, similar studies show data that do not meet the Hardy-Weinberg population laws [14,15,24]. Although the CYP3A4*1B/*1B genotype is overrepresented in prostate cancer patients, no differences were observed in the frequency of this allele between controls and cases (Table S4). Moreover, no association of this polymorphism with PSA, Gleason grade, or TNM was observed. On the other hand, it has been reported that the incidence rate of PCa is higher in patients with benign prostatic hyperplasia having the CYP3A4*1B allele compared to those having the CYP3A4*1A allele [25]. Considering the above, it would be interesting to assess whether this association is also present in the Mexican population. Another important consideration is the potential clinical impact of the CYP3A4*1B variant on the individual response to hormonal therapy.
A previous study reported a stronger association between the CYP3A4*1B genotype and Gleason grade and TNM score in patients diagnosed at the age of 65 years or older [14,15]. In the present patients population, the dominant model analysis showed that individuals diagnosed at the age of 65 years or older and with the CYP3A4*1B allele had a higher risk of a TNM$3 (OR = 2.2). However, this increase was not statistically significant. Moreover, the present study did not find an association between CYP3A4*1B polymorphisms and clinic characteristics of prostate cancer. Other studies have also failed to show an association between prostate cancer and this allelic variant [17]. The functional significance of the CYP3A4*1B polymorphism has been studied using in vitro and in vivo approaches. Transactivation studies indicate that the -290A/G variant results in an increase in reporter gene activity, suggesting that the CYP3A4*1B polymorphism is unlikely to decrease the capacity to inactivate testosterone and therefore  increase the risk for PCa development [26] [27]. More controversial results regarding in vivo studies have been reported. Wandel and collaborators [19] showed a modest difference in CYP3A4 activity between African Americans and European Americans associated with the CYP3A4*1B frequency. However, Lamba and collaborators [28] did not find any association between this polymorphism and the CYP3A4 phenotype. Ball [29] reported similar results. The lack of a correlation could be explained by other genetic polymorphisms involved in hormone metabolism, such as those presented at the CYP3A5 gene. Moreover, sequence variation of transcription factors that modulate the expression of CYP3A4 and other hormone metabolizing enzymes should be considered as well. In particular, PXR may play an important role, since several transporters and metabolizing enzymes, including CYP3A4, are under its transcriptional regulation. In the present study, neither PXR-HNF4/G nor PXR-HNF3b/T alleles were over represented in case group. However, a clear association between the PXR-HNF4/G allele and PSA levels was observed. We found that heterozygous and homozygous prostate cancer patients had a higher risk of presenting higher levels of PSA (OR = 2.46 and OR = 3.99, respectively). When the analysis was applied to patients diagnosed at the age of 65 years or older, the dominant model analysis showed that individuals with this allele presented an even higher risk of having high PSA levels $10 (OR = 10.8).
The PXR-HNF4 (69789A.G) polymorphism modifies an HNF4 binding site located in the PXR gene promoter, which results in lower PXR and CYP3A4 mRNA levels together with a decrease in CYP3A4 basal activity [30]. On the other hand, it has been shown that prostate tissue from prostate cancer patients express lower levels of PXR and CYP3A4 proteins compared to benign prostate tissue from control group [31] [10]. This most likely favors the conversion of testosterone to dihydrotestosterone (DHT) increasing the risk of developing PCa as well as prostate cancer markers such as PSA. PSA is a member of the kallikrein gene family. Several data indicate that PSA is under androgenic regulation. Induction of PSA mRNA is driven by mibolerone and DHT and inhibited by antiandrogens, suggesting that this effect is androgen receptor-dependent [32] [33] [34]. Once PSA is secreted it degrades extracellular proteins facilitating the invasion of prostate cancer cells [35]. PSA has been widely used to screen men for PCa leading to a dramatic reduction in PCa death rates. However, its use is controversial due to its limit to distinguish between cancer and benign prostatic hyperplasia, or between indolent and aggressive cancer resulting in overdetection and overtreatment. The present results should be understood by considering the above.
In addition to CYP3A4, PXR regulates the expression of other genes involved in steroid hormone metabolism, such as CYP17 and CYP24A1, which might account for alterations in testosterone bioavailability [36] [37]. The PXR-HNF4/G allele frequency in the control group was 0.37, which is similar to that observed for Caucasians (0.36) [30]. In contrast, this allele frequency in the prostate cancer group was 0.56. However, this difference was not statistically significant. Nevertheless, the presence of the PXR-HNF4/G allele is high in both populations, Caucasian and Mexicans, and therefore it would be interesting to explore possible associations of this polymorphism and drug metabolism as well as other pathologies.
Interestingly, the patients homozygous for the CYP3A4*1B allele were also homozygous for the PXR-HNF4/G allele. Therefore, we investigated whether an epistasis phenomenon was involved and found that the two alleles did not interact.
In summary, the present study is the first to show that the presence of the PXR-HNF4/G allele increases the risk of having higher levels of PSA in patients with prostate cancer. It is also the first study that evaluates, in a Mexican population, the association of CYP3A4 and PXR gene polymorphisms with PCa. In agreement with previous studies, the present data suggest that the CYP3A4*1B polymorphism appears to be a gene with low penetrance and therefore has moderate effects on the risk of developing prostate cancer. Finally, future studies should be performed to assess the PXR variants role in PCa development.