CYP3A43 Pro³⁴⁰Ala Polymorphism and Prostate Cancer Risk in African Americans and Caucasians

In the United States, prostate cancer is the most common form of cancer and is the second leading cause of cancer-related death among men (1). African American men have the world's highest incidence of prostate cancer and more than twice the death rate compared with Caucasian men; they also have significantly higher rate of disease severity at diagnosis (2-7). Epidemiologic evidence indicates that steroid hormones play a major role in prostate carcinogenesis and may partially explain the risk disparity between African Americans and Caucasians. Previous population-based studies have addressed the hypothesis that functional polymorphisms in genes involved in testosterone metabolism might be associated with the differences in prostate cancer risk among various ethnic populations; however, results have been inconsistent and contradictory (8-11).

Members of the cytochrome P450 (CYP) family of enzymes, which are responsible for >50% of drug metabolism, are heme-containing mono-oxygenases that catalyze hydroxylation of steroids (including testosterone, progesterone, and cortisol) and xenobiotics (12). The CYP3A family is composed of four known CYP3A genes in humans: CYP3A4, CYP3A5, CYP3A7, and CYP3A43. Each gene contains 13 exons and is located on chromosome 7q21-22.1, with CYP3A43 in an unusual head-to-head orientation with CYP3A4, CYP3A5, and CYP3A7 (13, 14). CYP3A43 shares DNA sequence homologies of 84%, 83%, and 82% and amino acid homologies of 76%, 76%, and 72% with CYP3A4, CYP3A5, and CYP3A7, respectively (13, 15).

Distribution of CYP3A43 mRNA in various tissues (liver, kidney, pancreas, prostate, etc.) by PCR amplification has shown that the expression level of CYP3A43 is considerably lower than CYP3A4, CYP3A5, and CYP3A7, although the level of expression of CYP3A43 does not necessarily reflect the extent of its actual role in xenobiotic metabolism (15). Westlind et al. found that CYP3A43 and CYP3A4 mRNA are expressed in prostate, colon, breast, lung, and pancreatic carcinoma; there is also some evidence of hepatic coexpression of all four mRNAs (12). Three different transcript variants are formed from alternate splicing of this gene. Interestingly, there are several hybrid CYP3A43/CYP3A4 and CYP3A43/CYP3A5 genes that have been formed by splicing CYP3A43 exon 1 to various CYP3A4 or CYP3A5 exons, resulting in hybrid mRNA products (13). The chimeric proteins formed from these transgenic splicing variants are not always enzymatically active; however, the longest chimeric isoform, encoded by the (1)_CYP3A43-(2-13)_CYP3A4 cDNA, can hydroxylate testosterone (13).

Xenobiotics and endogenous substances, such as steroid hormones, can activate the pregnane X receptor, a human orphan nuclear receptor, resulting in the induction of CYP3A genes. Two drugs known to induce CYP3A activity in primary hepatocytes are rifampicin and dexamethasone; however, the four CYP3A isoforms differed in expression levels when other drugs were used, indicating that the four isoforms are differentially regulated (14-16). In addition, there is very little sequence similarity among the 5′ untranslated regions of CYP3A43 and CYP3A4, CYP3A5, and CYP3A7 genes; regulatory elements in the 5′ region of CYP3A43 have not been found, although CYP3A4 has several known regulatory regions and binding sites in the 5′ untranslated region (17).

Several studies have addressed the hypothesis that functional polymorphisms in the CYP3A4 and CYP3A5 genes could be associated with prostate cancer risk among different ethnic populations. Rebbeck et al. found that an A-to-G mutation within the nifedipine-specific element, −293 bp from the transcription start site of the CYP3A4 gene, was associated with higher grade and stage in Caucasians with prostate cancer (18). Additional studies found evidence that African Americans possess the CYP3A4*1B (variant allele) at 6 to 10 times the frequency found in Caucasians; limited data support the hypothesis that the variant allele alters testosterone metabolism (19-21). The CYP3A5 6986 G > A (CYP3A5*3) variant correlates with function of the CYP3A5 enzyme and is in linkage disequilibrium with the CYP3A4 promoter variant and possibly other alleles (22). Plummer et al. found that the CYP3A4*1B/CYP3A5*3 haplotype (which is more common in African Americans than Caucasians) is positively associated with prostate cancer, but the CYP3A4*1B variant is inversely associated with risk among Caucasians with less aggressive disease (23).

Recently, three polymorphisms in the coding region of the CYP3A43 gene have been confirmed: a silent substitution in exon 11 (c.1047 C > T; designated CYP3A43*1B), a frameshift mutation in exon 2 (c.74delA; designated CYP3A43*2A), and a missense mutation in exon 10 (c.1018C > G; designated CYP3A43*3; ref. 24). Because the CYP3A43*3 genotype is most likely associated with changes in function, we investigated the association between CYP3A43*3 genotype and prostate cancer risk in a case-control study consisting of both African Americans and Caucasians.

Incident, histologically confirmed prostate cancer cases were recruited to the case-control study within 6 months of diagnosis starting in 1998 until 2003. Prostate cancer cases were recruited from the University of Arkansas for Medical Sciences, the University Hospital in Little Rock, the Central Arkansas Veteran's Health Care System in Little Rock, and the Jefferson Regional Medical Center in Pine Bluff, Arkansas. The recruitment rate for cases was 69% (Caucasians 76% and African Americans 55%) and for controls was 56% (Caucasians 58% and African Americans 55%). DNA was available for 99% of the cases and controls. Community controls were identified from Arkansas State Driver's License records, Centers for Medicare and Medicaid Services records, and a mass mailing database that covers ∼80% of Arkansas residents. Sixty-eight percent of the controls were from the mass mail database (Caucasians 62% and African Americans 71%), 14% of the controls were from a Centers for Medicare and Medicaid Services database (Caucasians 21% and African Americans 16%), and 19% of the controls were from the driver's license database (Caucasians 21% and African Americans 16%). Study controls were frequency matched to cases on age (±5 years) and race. Exclusion criteria for the case-control study included a history of cancer (except nonmelanoma skin cancer), uncontrolled cardiovascular disease, and hepatic dysfunction (as determined by bilirubin >1.5 mg/dL, aspartate aminotransferase >40 units/L, alkaline phosphatase >140 units/L, and abnormal renal function as determined by blood urea nitrogen >20 mg/dL and serum creatinine >1.8 mg/dL). Covariate data for the current study were obtained by conducting in-person interviews with cases and controls at the time of enrollment to the case-control study. Each study participant also provided a blood sample for DNA analyses. The appropriate institutional review board approvals were obtained for the study protocol. Signed informed consent was obtained for the interview, blood collection, and analyses of polymorphisms.

DNA was extracted from lymphocytes of study participants using a commercial kit (Qiagen, Inc., Valencia, CA) and the samples were genotyped for the CYP3A43*3 polymorphism at a commercial laboratory (BioServe Biotechnologies Ltd., Laurel, MD) by high-throughput, chip-based matrix-assisted laser desorption time-of-flight mass spectrometry (Sequenom, Inc., San Diego, CA) using the MassEXTEND reaction (Sequenom). PCR primers and extension primers were designed using SpectroDESIGNER software (Sequenom) and synthesized at BioServe Biotechnologies. Oligonucleotide sequences of the primers and probe were forward primer ACGTTGGATGCATTCTTGCTGAGGC, reverse primer ACGTTGGATGCCTGATGTCCAGCAGAAAC, and extension primer TCATCCCCTTACCTTATTGG.

All laboratory personnel were blinded to case-control status. Forty-four blinded duplicates were included in the genotype analyses and were 100% concordant.

The Wilcoxon rank-sum test was used for continuous variables to test that the distribution of subject characteristics was the same for cases and controls. The χ² test was used for categorical variables. Hardy-Weinberg equilibrium was assessed using χ² tests. Unconditional logistic regression was used to compute odds ratios (OR) and 95% confidence intervals to estimate relative risk of prostate cancer associated with genotypes. The variables used in the multivariate analyses were pack-years of smoking as a continuous variable, age at diagnosis for cases and at interview for controls as a continuous variable, and race as a categorical variable. Potential confounding of the association between genotype and cancer risk by potential risk factors was explored using Spearman rank correlation analyses and multivariate logistic regression models, including stepwise regression models both before and after stratification. If the potential confounder caused a significant change in the log likelihood estimate (P < 0.05) and a >20% change in the B coefficient, it was kept in the model for further multivariate analyses. Modification of the association between genotype and prostate cancer odds by smoking, age, and race (Ps for interactions) was examined by statistical tests of the first-order interaction term in the logistic regression models. Throughout the article, all Ps shown are two sided. All statistical analyses were done using the software package Stata (Stata Corp., College Station, TX).

Table 1 shows selected characteristics of the study population. The mean age of the control subjects was 62 years compared with 67 years in the prostate cancer patients (P < 0.05). Body mass index was the same comparing cases with controls. There were significantly more smokers in the case group than in the control group (P < 0.01). For the CYP3A43 genotype, the Hardy-Weinberg equilibrium assumption was tested in the control group and also stratified by race using χ² tests and we found no deviation from equilibrium. There was a significant difference (P < 0.008) in the distribution of CYP3A43 genotype comparing prostate cancer cases with controls, with the CYP3A43-Ala/Ala genotype being more prevalent in the cases compared with controls (5% and 2%, respectively).

Table 2 shows the association between CYP3A43 genotype and odds of prostate cancer. Genotype data were available for 465 prostate cancer patients and 467 control individuals. When all subjects were considered, there was a 3-fold increase in risk of prostate cancer among individuals with the CYP3A43-Ala/Ala genotype (OR, 3.03; 95% confidence interval, 1.27-7.23) compared with those with the CYP3A43-Pro/Pro genotype after adjusting for age, race, and pack-years of smoking. When tests for interaction were done, there was statistically significant interaction between CYP3A43 genotype and race (P < 0.001). Furthermore, we observed a borderline significant interaction between CYP3A43 genotype and smoking (P < 0.07).

Stratified analyses by race revealed that prevalence of the genotype categories were significantly different in African American controls compared with Caucasian controls (P < 0.001). Among African Americans, there was a 2.6-fold increase in risk of prostate cancer among individuals possessing the CYP3A43-Ala/Ala genotype (OR, 2.64; 95% confidence interval, 1.01-7.00) compared with those with the CYP3A43-Pro/Pro genotype after adjusting for age and pack-years of smoking. Among the Caucasians, there were only three individuals with the CYP3A43-Ala/Ala genotype and all individuals were prostate cancer cases. Stratified analyses by smoking status revealed that there was a statistically significant association between CYP3A43-Ala/Ala genotype and prostate cancer odds among smokers (OR, 8.55; 95% confidence interval, 1.06-68.99) but not among the nonsmokers. The limitation of this finding is that the 95% confidence interval for the association is quite wide.

This study shows an association between prostate cancer risk in African Americans and a newly discovered polymorphism of the CYP3A43 gene that encodes for an enzyme involved in steroid hormone metabolism. The nonconservative substitution of proline, an α-imino acid, with alanine, an amino acid, could have implications for structural integrity or less efficient folding of the protein, thereby reflecting differences in specific activity between the wild-type and the mutant forms. Glykos et al. showed that a single alanine-to-proline substitution was sufficient for changing the topology of a small protein by changing the “surface properties of the protein and the packing of its hydrophobic core while retaining some of the biological activity of the wild-type molecule” (25). One known function of the CYP3A43 wild-type protein is conversion of free testosterone to 6-β-hydroxytestosterone, resulting in testosterone inactivation (15). We hypothesize that the variant allele (Ala/Ala) decreases CYP3A43 protein activity, thereby reducing 6-β-hydroxytestosterone oxidation; therefore, more circulating testosterone is available for conversion to its active form, dihydrotestosterone. Some studies have found a correlation between prostate cancer risk and elevated circulating testosterone levels in high-risk African American men (26-30). However, functional studies on both wild-type and variant forms of the CYP3A43 enzyme are necessary to elucidate the biological mechanisms contributing to the increased risk of prostate cancer found in this study. In addition, other functional studies of substrates in both androgen and estrogen pathways need to be investigated, because CYP3A43 seems to differ structurally from the other CYP3A family members in substrate recognition sites SRS-1 to SRS-6 (31).

Cigarette smoking is one putative environmental risk factors for prostate cancer; cigarette smoking has been associated with higher levels of testosterone and lower levels of estradiol in men (32). Because testosterone can enhance cell proliferation in the prostate, and estrogens can reduce testicular androgen production, the necessary hormonal environment may be produced to promote prostate cancer initiation (33). In a population-based, case-control study with Caucasians and African Americans, Plaskon et al. found that smoking status was a moderate risk factor for prostate cancer (33). In addition, this study found that the relative risk significantly increased with >40 years duration and with >40 pack-years of exposure. In our study population, cigarette smoking is also a risk factor for prostate cancer (Table 1). However, the majority of studies conducted to date suggest that smoking is not associated with increased risk of prostate cancer (34-36). On the other hand, smoking has been noted to increase prostate cancer mortality (37, 38). We found a statistically significant association between CYP3A43-Ala/Ala genotype and prostate cancer odds among smokers (OR, 8.55; 95% confidence interval, 1.06-68.99) after adjusting for age and pack-years of smoking; however, the 95% confidence interval is quite wide. Nevertheless, this study needs to be replicated in much larger population-based studies of prostate cancer with racially diverse populations because the prevalence of the CYP3A43-Ala/Ala genotype is <10% in both Caucasians and African Americans.

Study results may be affected by sources of bias that may affect case-control studies, including recall, misclassification, and selection bias. The in-person interview and in-depth assessment of histories is likely to reduce recall bias to some extent as well as the fact that cases were interviewed as soon after diagnosis as possible. Regarding misclassification, we have confidence in classification of genotypes that were in Hardy-Weinberg equilibrium and were identified using the very accurate matrix-assisted laser desorption time-of-flight technology. Although there is always the possibility that controls could have undiagnosed prostate cancer, we conducted prostate-specific antigen assays on all control participants and excluded all men whose values were >4 ng/mL. It is more difficult to exclude the possibility of selection bias in case-control studies, although there should be no differential participation based on genotypes; thus, selection bias should not affect the main effects of the genetic variants. Because of the small proportion of men homozygous for the variant allele, larger studies with more African American and Caucasian men are needed; however, this is one of the first studies to evaluate CYP3A43, a gene likely to be important in testosterone metabolism.

Our laboratory is in the process of measuring plasma levels of circulating hormones so that we can evaluate the association between CYP3A43*3 genotype and steroid hormones. For example, levels of testosterone, its precursor androsterone, and its major metabolite dihydrotestosterone vary considerably when stratifying by race, age, and smoking status. In the population-based case-control study, Plaskon et al. (33) measured serum total testosterone and sex hormone binding globulin levels in current smokers and nonsmokers in the control group; they found significantly higher levels of testosterone and sex hormone binding globulin in current smokers.

Because several CYP3A genes play an essential role in steroid and drug metabolism, are candidate genes for prostate cancer risk, and are in close proximity to the CYP3A43 gene, linkage disequilibrium studies need to be done between the CYP3A43 Pro³⁴⁰Ala variant and known functional allelic variants in the CYP3A4, CYP3A5, and CYP3A7 genes. Prostate cancer risk may not be directly associated with the CYP3A43 gene but rather the CYP3A43 gene may be in linkage disequilibrium with other member(s) of the CYP3A family that are the actual causative agent(s). For example, Zeigler-Johnson et al. did not find any association between CYP3A5 known functional variants and prostate cancer risk; however, when CYP3A4/CYP3A5 haplotypes were evaluated, an inverse relationship was found with prostate cancer (39). Comprehensive analysis of the relationship between the CYP3A family and prostate cancer risk is further complicated by the high frequency of alternatively spliced CYP3A43 mRNA variants, which may generate protein isoforms that function differently from the canonical protein.

In conclusion, we report one of the first case-control studies to associate polymorphisms in the CYP3A43 gene with prostate cancer susceptibility. Our results indicate that the CYP3A43 Pro³⁴⁰Ala polymorphism prevalence differs by race and contributes to prostate cancer risk in African Americans. Because African Americans have the highest incidence and mortality rates from prostate cancer in the world, the study of polymorphisms that differ in prevalence by race, such as the CYP3A43 Pro³⁴⁰Ala polymorphism, provide an opportunity to foster insights into prostate cancer risk disparities. A key focus area in our research group is the study of these polymorphisms and prostate cancer susceptibility. Haplotype and linkage analyses of polymorphisms in the CYP3A4, CYP3A5, and CYP3A43 genes may further elucidate relationships between variants in the CYP3A family and prostate cancer risk. Studies are also under way in our laboratory to elucidate the functional significance of the CYP3A43 Pro³⁴⁰Ala change.

We thank Joni Dienstag for assistance during the study.