CYP3A4 Polymorphisms—Potential Risk Factors for Breast and Prostate Cancer: A HuGE Review
Channa Keshava¹, Erin C. McCanlies² and Ainsley Weston¹

¹ Molecular Epidemiology Team, Toxicology and Molecular Biology Branch, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, WV.
² Biostatistics and Epidemiology Branch, Health Effects Laboratory Division, National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Morgantown, WV.

Received for publication February 9, 2004. Accepted for publication May 21, 2004.

The steroid hydroxylase CYP3A4 is the most abundant P-450 enzyme in the human liver, and CYP3A enzymes metabolize more than 50% of prescription drugs. The CYP3A4 gene is expressed in the liver, gut, colon, prostate, and breast. Individual variation in CYP3A4 may play a role in breast and prostate carcinogenesis through modulation of sex hormone metabolite levels. Alternatively, CYP3A4 can metabolically activate exogenous carcinogens. CYP3A4 activity varies widely in humans, and more than 78 DNA sequence polymorphisms are known. These observations prompted the hypothesis that variant CYP3A4 may be involved in breast and prostate cancer. Two epidemiologic studies of breast cancer and five of prostate cancer examined CYP3A4 genotypes. A US study showed that inheritance of CYP3A4*1B correlates with early menarche, a breast cancer risk factor. However, an Australian breast cancer case-control study found no association with CYP3A4*1B. Two Scottish prospective studies showed CYP3A4*1B to be a risk factor for prostate cancer among men with benign prostatic hyperplasia. Three other studies were undertaken in the United States: two were case-only studies and the other was a case-sibling control study. Although results for African Americans were inconsistent, these studies suggested that CYP3A4*1B was associated with markers of advanced disease. These observations support the notion that development of robust, conventional molecular epidemiologic case-control studies to address these questions, including gene-gene and gene-environment interactions, will be timely.

Cytochrome P-450 genes constitute a superfamily coding for different isozymes that account for phase I drug metabolism. In humans, more than 40 different cytochrome P-450s have been identified and sequenced (1-3). CYP3A4 was originally named nifedipine oxidase for its ability to metabolize the antianginal drug nifedipine (Figure 1). However, since cytochrome P-450 substrate specificities proved to be broad and overlapping, a formal evolutionary nomenclature was defined that more aptly categorizes each isozyme. CYP denotes cytochrome P-450 for humans (cyp for mouse). The gene families are then designated by numbers following CYP. Originally, these were Roman numerals but were changed to Arabic. Subfamilies are represented by a letter followed by a number for the individual gene (4,5). For example, for CYP3A4, 3 denotes the gene family, A the subfamily, and 4 the gene coding for a specific polypeptide. Current estimates indicate that each mammalian species may have between 40 and 200 distinct functional cytochrome P-450 genes (4,6).

click to view

Figure 1  The chemical structure of nifedipine.

The CYP3A4 gene, located on chromosome 7q21.3-q22.1, is 27,592 base pairs long and has 13 exons. The promoter region encompasses a basal transcription element (–35 to –50). Also present in the 5' untranslated region (UTR) are an AP-3 binding site, a p53 binding motif (a specific DNA sequence to which the protein p53 can attach), a hepatocyte nuclear factor-4 element, two hepatocyte nuclear factor-5 elements, a glucocorticoid response element, and an estrogen response element (7-8). The corresponding protein is also known as CYPIIIA4/nifedipine oxidase/NF-25/P-450-PCN1 and consists of 502 amino acids with a molecular weight of 57.29 kDa. It is membrane bound and is present in endoplasmic reticulum.

CYP3A4 is expressed in the prostate, breast, gut, colon, and small intestine, but its expression is most abundant in the human liver, accounting for 30 percent of the total CYP protein content (9-14). It exhibits a broad substrate specificity and is responsible for oxidation of many therapeutic drugs and a variety of structurally unrelated compounds, including steroids, fatty acids, and xenobiotics. In liver microsomes, it is involved in a nicotinamide adenine dinucleotide phosphate-dependent electron transport pathway. Pertinent to this review, CYP3A4 has an important role in the oxidation of both testosterone (2ß-, 6ß-, or 15ß-hydroxytestosterone) and estrogen (4- and 16-hydroxylation) (15-18). It can be induced by various compounds including drugs, pesticides, and carcinogens, resulting in high CYP3A4 levels in liver and other tissues, including mammary. A mutation in CYP3A4 may lead to a reduced potential for oxidizing testosterone, leaving a greater bioavailability of the hormone to be metabolized intracellularly to its biologically active form of dihydroxytestosterone, the principal androgenic hormone involved in regulating prostate growth (Figure 2).

click to view

Figure 2  CYP3A4 and androgen metabolism. CYP3A4 is associated with oxidative  deactivation of testosterone and is responsible for regulation of  testosterone’s metabolism to 2ß-, 6ß-, and 15ß-hydroxytestosterone, less  biologically active forms of testosterone. SRD5A2, steroid-5- -reductase  gene; AR, androgen receptor gene.

Prevalence
Ovid software (Ovid Technologies, Inc., New York, New York) was used to search the MEDLINE database (National Library of Medicine, Bethesda, Maryland) between January 1, 1989, and May 3, 2004, using the search terms cytochrome P-450, breast neoplasms, and prostate neoplasms. The GenBank resource at the National Center for Biotechnology Information of the National Institutes of Health (Bethesda, Maryland) was used to search for DNA sequence information. Accession numbers for gene sequences are those of the National Center for Biotechnology Information. Numerical locations of polymorphisms are given as distances in base pairs from the methionine translational start signal (ATG), where A is nucleotide +1 and its adjacent 3' nucleotide is –1. The A in the ATG site for CYP3A4 can be found at position number 62037 in the GenBank entry identified by accession number AF280107.

Allelic frequencies and their 95 percent confidence intervals were determined according to standard methods by using Microsoft Excel software (19). Allele frequencies were used to calculate the expected numbers of persons with a given genotype according to Hardy-Weinberg population laws. Chi-square statistics were used to determine whether experimental observations departed significantly from Hardy-Weinberg equilibria (20). Crude odds ratios were calculated according to the Mantel-Haenszel method (21) by using SAS statistical software (22).

Discovery of CYP3A4 gene variants
Although human interindividual variation in 6ß-hydroxycortisol:cortisol ratios suggested their existence, genetic polymorphisms of CYP3A4 were unknown until 1996 (23). Since 1998, at least 78 nucleotide sequence variations of CYP3A4 have been identified (Tables 1, 2, and 3). Most of the variants are correctly characterized as polymorphisms (frequency > 0.01, at least 1 percent of the chromosomes in a given population), whereas many are technically constitutive mutations (frequency < 0.01). Some genetic variants are named according to the scheme outlined above; some have trivial names. However, it is important to recognize that haplotypes comprising multiple polymorphisms, or mutations, have barely been recognized. Originally, three pairwise haplotypes were described: CYP3A4*15A, CYP3A4*15B, and CYP3A4*19. Most recently, a Japanese consortium released details of 25 haplotypes, including 17 novel DNA-sequence variants (24). Note that several of these haplotypes are not yet unambiguous, and some may be limited to this population. In addition, a recent study described haplotypes between CYP3A4*1B and CYP3A5*1 as well as CYP3A4*1B and CYP3A5*3 in prostate cancer (25).

click to view

Table 1  DNA sequence variants of CYP3A4 found in the enhancer region, promoter  region, and 3'UTR*

click to view

Table 2   Findings from studies of the relation between polymorphisms in the     5'-  untranslated region of the cytochrome P-450c17 (CYP17) gene and   hormone levels in men

click to view

Table 3   Findings from studies of the relation between polymorphisms in the     5'-  untranslated region of the cytochrome P-450c17 (CYP17) gene and   hormone levels in men

study conducted at the Hospital of the University of Pennsylvania, Philadelphia, specifically to identify CYP3A4 polymorphisms evaluated a panel of DNA samples assembled from buccal swabs of 94 Caucasian male volunteers with no history of cancer (26). A single nucleotide polymorphism was identified in a 5' regulatory element of CYP3A4 by using conformation sensitive gel electrophoresis and direct DNA sequence analysis. This single nucleotide polymorphism (A/G) at nucleotide position –392 is found in the nifedipine promoter region. The common variant was designated CYP3A4*1A, and the newly discovered minor variant was designated CYP3A4*1B (Table 1). This polymorphism was independently identified by direct DNA sequencing and gel mobility shift assay of liver DNA obtained through either organ donors or normal portions of surgical specimens (27).

Further confirmation of the CYP3A4*1B variant was made when sequence determination for the 5' flanking region and the entire coding region was performed for 20 Caucasians, 20 African Americans, and 20 Chinese (28). This strategy ensures a greater than 98.5 percent probability of finding any polymorphism that has a frequency of at least 0.10, as determined by using binomial probability (29). Consequently, this approach revealed several additional CYP3A4 polymorphisms. At codon 222, an amino acid substitution serine/proline was observed (CYP3A4*2). Another rare allelic variant in codon 455, designated CYP3A4*3, was found in a single Chinese subject. A silent polymorphism was observed in African Americans; this polymorphism was apparently confirmed by Lamba et al. (30) by using direct DNA sequencing. These authors designated this polymorphism CYP3A4^I193I, denoting a silent change in an isoleucine codon (Table 2).

When polymerase chain reaction (PCR)-restriction fragment length polymorphism analysis was used, three more novel variants of CYP3A4 were found in Chinese subjects (n = 102) (31). These alleles were designated CYP3A4*4, CYP3A4*5, and CYP3A4*6. When PCR with standard fluorescence-based DNA sequencing and DNA from a mixed population originating in a variety of institutions in the United States and the United Kingdom (32) was used, two more allelic variants in the promoter region designated CYP3A4*1C and CYP3A4*1D were found. Similarly, but by using a single stranded conformation polymorphism in addition to DNA sequencing in a geographically diverse population of Caucasians, Hamzeiy et al. (33) reported two additional promoter-region single nucleotide polymorphisms, CYP3A4*1E and CYP3A4*1F. This group also reported a nine nucleotide insertion between positions –844 and –845 in the CYP3A4*15A allele, subsequently designated CYP3A4*15B (Table 1).

Seventeen more genetic variants were identified by Eiselt et al. (34). Seven resulted in amino acid substitutions, seven were in introns, two were silent in exons, and one was in the 3'UTR. All variants were single nucleotide polymorphisms discovered by PCR and DNA sequencing in samples originating from European and Middle Eastern populations. The amino acid substitutions were designated CYP3A4*7, CYP3A4*8, CYP3A4*9, CYP3A4*10, CYP3A4*11, CYP3A4*12, and CYP3A4*13 (Table 2). Silent DNA-sequence variants, and those in introns and the 3'UTR, were designated M10–M19 (Tables 1, 2, and 3).

Using direct sequencing of genomic DNA from multiple ethnic groups, Dai et al. (35) identified 28 polymorphisms including two new ones in the 3'UTR (Table 1) and three new coding-region polymorphisms, CYP3A4*17, CYP3A4*18, and CYP3A4*19 (Table 2 ). They also described two silent coding-region polymorphisms, one in exon 10 and one in exon 11 (Table 2), as well as seven new polymorphisms in introns 4, 5, 7, and 9–11 (Table 3). Note that CYP3A4*19 describes the haplotype of the single nucleotide polymorphisms at positions 85273 and 82266 (Tables 2 and 3, respectively).

Lamba et al. (30) used small groups of ethnically diverse samples (numbering between five and 10), as well as 21 African-American and 53 Caucasian subjects, to investigate the entire gene by PCR-DNA sequencing. In addition to the polymorphisms described above, they presented data for 20 additional polymorphisms or mutations: five were in the 5'UTR/promoter region, three were in the 3'UTR, and three were in coding regions (Tables 1 and 2). The rest were found in introns 2–4, 7, and 10 (Table 3). As of this writing (May 3, 2004), all of the known 78 individual, constitutive DNA sequence variants (inherited polymorphisms or mutations) of CYP3A4 are listed in (Tables 1, 2, and 3).

Genotypes and allele frequencies
Genotypes and allelic frequencies for CYP3A4*1B, the minor variant of the original CYP3A4 promoter-region polymorphism, are given in Table 4. The frequency of this allele varies greatly between different ethnic/racial populations. It has been found to be the major variant among people of African origin, but six studies have failed to find this variant in Chinese, Taiwanese, or Japanese (24, 28, 36-39). The allelic frequency of CYP3A4*1B has been found to be 0.036–0.096 in Caucasians, 0.480–0.800 in African Americans, 0.093–0.107 in Hispanics, 0.089 in Saudis, and 0.690 in Ghanaians (26, 27, 40-43) (Table 4). However, one small study of 15 European Americans did not find the minor variant (44).

click to view

Table 4   Findings from studies of the relation between polymorphisms in the     5'-  untranslated region of the cytochrome P-450c17 (CYP17) gene and   hormone levels in men

The allelic frequency of CYP3A4*1B has been the most extensively studied (Table 4). Fourteen studies examined CYP3A4*1B variants in nondiseased or healthy Caucasian populations; in five of these studies, the genotypic distribution did not conform to Hardy-Weinberg population laws (26, 36, 39, 40, 45). In each case, the reason for this finding was an excess number of CYP3A4*1B variant homozygotes. There could be several reasons, including faulty genotyping; however, the underlying basis is obscure and is not discussed in the original articles. In two of three prostate cancer populations for which the Hardy-Weinberg chi-square statistic could be calculated, highly significant deviation from Hardy-Weinberg equilibrium was observed (Table 4). In the African-American population, a large excess of CYP3A4*1B variant homozygotes was observed. In the Caucasian population, although no homozygotes were observed, there was an excess of heterozygotes. When a specific allele is associated with disease, as may be the case with the CYP3A4*1B variant, deviation from Hardy-Weinberg equilibrium among cases is not unexpected.

Almost all of the other DNA-sequence variants described fall into the category of rare polymorphisms (frequency = 0.01–0.03) or mutations (frequency < 0.01). Many of the populations in which these variants were found were inadequate to properly determine genotypic and allelic frequencies (Tables 5 and 6). Examples of inadequate populations for these purposes are those of mixed descent (33) or those with very small sample sizes (30). The frequencies of nonsynonymous coding region CYP3A4 DNA-sequence variants are given in Table 5 . There are 22, of which nine are technically mutations. In addition, confirmation in a robust, independent population is needed for most of these alleles (24, 28, 30, 31, 34, 35).

click to view

Table 5   Findings from studies of the relation between polymorphisms in the     5'-  untranslated region of the cytochrome P-450c17 (CYP17) gene and   hormone levels in men

click to view

Table 6   Findings from studies of the relation between polymorphisms in the     5'-  untranslated region of the cytochrome P-450c17 (CYP17) gene and   hormone levels in men

Several non-coding-region polymorphisms were reported with their allelic frequencies, but genotyping data were generally not available. In two studies of African Americans (n = 24 and n = 21), nine polymorphisms of moderately high frequency were found in intron 2 (T/C, 5917, 0.146), exon 7 (CYP3A4^I193I, 0.05; also previously reported to be 0.046 by Sata et al. (28)), intron 7 (T/G, 15753, 0.48 and 0.50; T/C, 15783, 0.087 and 0.05; T/A, 15837, 0.065), intron 10 (G/A, 20230, 0.73 and 0.50; C/T, 20265, 0.083; G/C, 20309, 0.104 and 0.05), and intron 11 (C/T, 23081, 0.21 and 0.15) (Table 6) (30, 35). Among 53 Caucasians, two polymorphisms of moderately high frequency were found in intron 7 (T/G, 15753, 0.065) and intron 10 (G/A, 20230, 0.11) (30). The frequency of the intron 10 polymorphism in Caucasians was also reported to be 0.095 and 0.146 by Eiselt et al. (34) and Dai et al. (35), respectively. The frequency of this polymorphism in Asians was reported to be 0.375 (35).

A recent report (24) described the frequency of 24 CYP3A4 DNA-sequence variants among 416 Japanese using PCR-DNA sequencing. Seven were already known (76236, 79698ˆ9, 82266, CYP3A4*6, *11, *16, and *18), and most were mutations (the frequency of 20 of the variants was <0.01, 95 percent confidence interval (CI): 0, 0.008). In addition, another recent report described a TGT insertion in CLEM 4, an enhancer of CYP3A4 (46). This TGT insertion was described as being located approximately 11 kilobases to the 5' of the CYP3A4 ATG site.

Breast cancer is the second leading cause of death in women. There were 211,300 new cases of breast cancer in the United States in 2003 and 40,000 deaths from the disease (47). Several established risk factors for breast cancer include family history of breast cancer, early menarche, late age at first birth and nulliparity, and genetic factors (BRCA1 and BRCA2). Other factors have been strongly implicated but are less well established, including oral contraceptive use, hormone replacement therapy, low physical exercise levels, and obesity (48). Most of these factors point toward increased exposure to estrogen. Several epidemiologic studies have shown that allelic variation in other genes such as p53 may render a woman susceptible to breast cancer (49-53). Few studies have investigated a role for CYP3A4 in breast cancer risk.

Prostate cancer is the most common non-skin-related cancer affecting men in the United States and the second leading cause of cancer-related deaths (54). An estimated 220,900 new cancer cases and an estimated 28,900 deaths were expected for 2003 (47). Major risk factors for prostate cancer are less well defined than those for breast cancer. There is compelling evidence of family history for prostate cancer (55); other major risk factors include race (the disease is common in Caucasians and African Americans but not Africans, rare in Asians) and age. Other possible risk factors include diet, multiple sexual partners, urban environment, body mass index, physical activity, vasectomy, and hormonal factors (56). These observations imply that different lifestyle factors are important, and, given that factors that increase estrogen levels and reduce testosterone levels are protective, it is probable that exposure to tes-tosterone is important.

Several genes have been identified as accounting for some of the approximately 9 percent of prostate cancers that are familial. RANSEL is an interferon-inducible endoribonuclease linked to HPC1, which is also associated with familial prostate cancer. Whereas mutations have been identified for RANSEL, HPC1 was found as the result of a genome-wide scan. Other genes with prostate-cancer-associated mutations include MSR1 and ELAC2 (56). Polymorphisms in canonical repeat regions of both AR and SRD5A2 also are thought to increase prostate cancer susceptibility, and a CYP17 (steroid 17-lyase) polymorphism may also be involved (Figure 2) (57-64).

Interindividual and interethnic variability in the expression of CYP3A4 accounts for large differences in disposition to xenobiotics, therapeutic drugs, and endobiotics (steroids) such as the oxidation of testosterone and the hydroxylation of estrogens. This difference in disposition to xenobiotics and drugs has been found to be at least 10–40-fold (9, 14, 27, 30, 65-69). Interestingly, population distributions of both rates of steroid hydroxylation (6ß-hydroxylation of tes-tosterone pmol/mg per minute) and metabolite:substrate ratio (e.g., 6ß-hydroxycortisol/cortisol) appear to be unimodal (69). This study was performed before CYP3A4 polymorphisms were discovered, but several studies have addressed the question of CYP3A4 polymorphism-activity relations (27, 28, 31, 34, 35, 38, 40, 44). Two in vivo studies, using erythromycin and midazolam as probe drugs, failed to find any activity relations with CYP3A4*1B, *2, *4, *5, *6, *8, *11, *12, or *13 (38, 40). A third study using midazolam noted a modest reduction in clearance (30 percent, p = 0.02) associated with CYP3A4*1B (44). Two in vitro studies of testosterone hydroxylase found no activity relations with CYP3A4*1B, *3, *7, *8, *9, *10, *11, *12, or *13 (27, 31). A third study, which additionally used clorpyrifos, found decreased activity with CYP3A4*17, increased activity with CYP3A4*18, and no associations with CYP3A4*3 or *19 (35). A fourth study, which also used nifedipine, found altered enzyme kinetics associated with CYP3A4*2 (higher K_m and lower V_max) but no overall intrinsic difference in clearance characteristics from CYP3A4*1A (28). Taken together, these data suggest a lack of evidence to date that the major polymorphic variants in CYP3A4 have any association with CYP3A4 activity.

With this information in mind, it does not logically follow that CYP3A4 polymorphisms would be associated with steroid metabolism related to breast and prostate cancer. However, the original polymorphism association study of prostate cancer was published before the phenotyping studies (26). Moreover, the rate of prostate cancer in African Americans (230/105) is higher than that in Caucasians (150/105), which is higher than the rate in Asians (80/105) (Figure 3). This trend roughly correlates with the frequency of the promoter-region CYP3A4*1B variant in these populations (Table 4), potentially accounting for the fact that African-American men present with more severe forms of prostate cancer, possibly leading to a more aggressive course of the disease (70, 71).

click to view

Figure 3  Pathway of steroid hormone synthesis and metabolism, including the role  of the cytochrome P-450c17 (CYP17) gene.

Breast cancer
CYP3A4 plays a major role in the 4- and 16-hydroxylation of estrogens, particularly estrone, the predominant form of estrogens in postmenopausal women (16, 72-76). This enzyme, in conjunction with the action of CYP1A2, which catalyzes the formation of 2-hydroxyestrone, may determine 2-hydroxyestrone:16-hydroxyestrone ratios that in turn may be associated with breast carcinogenesis (77, 78).

Evidence suggests that overexpression of CYP3A4 may be associated with breast cancer. To determine which members of the cytochrome P-450 superfamily are expressed in human breast tissue and tumors, Huang et al. (13) studied the mRNA expression levels. CYP3A4 mRNA was present in 70 percent of normal breast tissues and 18 percent of tumor tissues; however, the sample size was very small (13 for normal tissue, 11 for tumor tissue).

Zheng et al. (79) evaluated the association between urinary cortisol ratios and breast cancer risk in a subgroup of women who participated in a population-based case-control study in Shanghai, China. They found a strong association between the risk of breast cancer and urinary 6ß-hydroxycortisol:cortisol ratios, where the metabolite:steroid ratio is a surrogate measure of CYP3A4 activity. They further showed that the risk increased in a dose-response manner. This association appeared to be driven by cortisol metabolism in older women (Figure 4). Taken together, these investigations provided some impetus for recent genotyping studies of CYP3A4 in relation to breast cancer risk.

click to view

Figure 4  Pathway of steroid hormone synthesis and metabolism, including the role  of the cytochrome P-450c17 (CYP17) gene.

There are two known genotyping studies of CYP3A4 and breast cancer or breast carcinogenesis (80, 81): one is a molecular epidemiologic case-control study, but the other links genotype and breast cancer risk factors in a small group of healthy, young women. Details of these studies are given in (Tables 7 and 8) . In the case-control study of Australian, Caucasian women, CYP3A4*1B was not associated with breast cancer (odds ratio (OR) = 0.86, 95 percent CI: 0.54, 1.33) or ovarian cancer (OR = 1.51, 95 percent CI: 0.80, 2.89) (80) (Table 7). This conclusion did not change when the data were stratified by age (<40/>40 years) or menopausal status. However, in a group of US girls (n = 137; 39 African American, 57 Hispanic, and 41 Caucasian) early-onset menarche, a breast cancer risk factor, was associated with inheritance of the CYP3A4*1B allele (odds trend = 3.21, 95 percent CI: 1.62, 6.89) (81).

click to view

Table 7   Findings from studies of the relation between polymorphisms in the     5'-  untranslated region of the cytochrome P-450c17 (CYP17) gene and   hormone levels in men

click to view

Table 8   Findings from studies of the relation between polymorphisms in the     5'-  untranslated region of the cytochrome P-450c17 (CYP17) gene and   hormone levels in men

There is evidence for a correlation between CYP3A4*1B and early life events associated with breast cancer risk, but one case-control study failed to find an association between breast cancer and CYP3A4*1B (80). These data suggest that, to address this question further, future case-control study designs should specifically and rigorously include issues of estrogen exposure to tease out what might be a small, but significant risk factor.

Prostate cancer
Differential steroid metabolite levels may result from CYP3A4 polymorphisms, and CYP3A4 is involved in the oxidation of testosterone to 2ß-, 6ß-, or 15ß-hydroxytestosterone. Potential interindividual variation in CYP3A4 function may play a role in androgen-mediated prostate carcinogenesis if the bioavailability of testosterone is affected, because steroid 5-reductase converts testosterone to dihydrotestosterone, which mediates prostate cell growth (Figure 2).

To our knowledge, there are five molecular epidemiologic studies of prostate cancer (25, 26, 39, 82, 83). Two are case-only studies linking clinical characteristics of disease with genotype, two are prospective studies of high-risk men (benign prostatic hyperplasia cases), and one is a case-sibling control study. Details of these studies, including genotyping methods used, are given in (Tables 7 and 8) .

The first genotyping study known to approach this question included the discovery of the promoter-region variant CYP3A4*1B (26). In this case-only study, association between a variety of clinical characteristics of prostate cancer and inheritance of CYP3A4*1B was investigated. For 230 incident, non-Hispanic, Caucasian prostate cancer cases recruited in the United States, the sequence context in the region of the nifedipine-specific responsive element was investigated by using a PCR-conformation sensitive gel electrophoresis method. Although the method was capable of distinguishing heterozygotes from homozygotes, all carriers of the CYP3A4*1B allele were collapsed into a single group, likely because homozygotes were somewhat infrequent. Here, we compared the CYP3A4*1B carrier frequencies between the 230 cases and the 94 healthy, unrelated Caucasians that Rebbeck et al. (26) referred to as a "reference panel" and found no risk of prostate cancer associated with carrier status (OR = 1.17, 95 percent CI: 0.62, 2.24) (Table 7). Furthermore, the Hardy-Weinberg chi-square statistic for the putative control group was highly significant (² = 6.49, p = 0.011). Therefore, it is possible that this comparison is invalid, although the mean age of each group was similar (63.3 vs. 63.4 years).

In the context of the case-only study, the focus was on comparison of CYP3A4*1B carrier status with the clinical attributes of prostate cancer, including age at diagnosis, prostate-specific antigen status, combined tumor grade (Gleason) and tumor-lymph node-metastasis, and family history (26). These analyses revealed that carriers of CYP3A4*1B were more likely to have tumors of a higher stage and grade. This effect was found to be greatest in older patients (diagnosed after 63 years of age) and in patients without a family history of the disease. The relative risk of advanced tumor stage associated with inheritance of CYP3A4*1B, when adjusting for detection method and age in a logistic regression model, was 2.1 (95 percent CI: 1.1, 4.1). The adjusted odds ratio for those patients with no family history of prostate cancer increased to 2.7 (95 percent CI: 1.2, 5.6). For those patients diagnosed at a later age (>63 years), the odds ratio increased to 6.7 (95 percent CI: 2.5, 17.7); for the older patients (>63 years of age) with no such family history, it was 9.5 (95 percent CI: 2.5, 35.2). No association was found between inheritance of CYP3A4*1B and prostate-specific antigen at the time of diagnosis, and there was no association with family history of cancer.

Subsequently, a functional role for CYP3A4*1B was investigated, and initial data suggested that higher levels of CYP3A4 expression are associated with it versus the CYP3A4*1A allele (37, 42). However, later mechanistic studies have not supported this link (83). In other studies, although amounts of CYP3A4 protein in liver microsomes have been observed to vary by race, no single CYP3A4 allele has been significantly linked with the CYP3A4 phenotype (9, 27, 28, 31, 34, 35, 38, 40, 83).

Paris et al. (39) evaluated the CYP3A4*1B genotype frequencies in 174 African-American prostate cancer cases and 116 healthy volunteers (Table 7). When these groups were compared, the crude odds ratio for CYP3A4*1B carriers was not significant (OR = 1.1, 95 percent CI: 0.6, 2.1). However, carriers of CYP3A4*1B homozygotes were at a significantly increased risk (OR = 2.2, 95 percent CI: 1.4, 3.7; (Table 4 and 7). This result was also reflected by the fact that an excess of homozygotes was detected in the Hardy-Weinberg statistic (² = 6.8, p < 0.01), and it was consistent with the initial study of US Caucasians (26). When a case-only design was used to examine the impact of inheriting CYP3A4*1B, it was found to be associated with advanced clinical characteristics. Men with CYP3A4*1B variant homozygotes were also more likely to present with a higher grade and stage of prostate cancer (OR = 1.7, 95 percent CI: 0.9, 3.4), and the association was even stronger for older men (>65 years of age) (OR = 2.4, 95 percent CI: 1.1, 5.4). These results are broadly consistent with those of Rebbeck et al. (26).

A CYP3A4*1B genotyping study was conducted in a group of 84 Scottish, Caucasian men; all had a diagnosis of benign prostatic hyperplasia and had been recruited prospectively with respect to a prostate cancer diagnosis (82). In this group, inheritance of CYP3A4*1B was associated with a greater than sixfold increase in risk (relative risk = 6.3, 95 percent CI: 2.3, 17.3; Table 7 of developing prostate cancer over a period of 6–15 years. No CYP3A4*1B homozygotes were observed in this study group; however, heterozygotes occurred in excess in the case group (Hardy-Weinberg ² = 4.8, p = 0.02). In a second cohort study of 400 men with benign prostatic hyperplasia, for 21 men who developed prostate cancer, the relative risk associated with inheritance of CYP3A4*1B was 2.7 (95 percent CI: 0.77, 7.66) (Table 7) (83).

A study of Caucasians (n = 816) and African Americans (n = 76), in which a family-based case-control design was used (sibling pairs: 440 cases and 480 controls), investigated the CYP3A4*1B and CYP3A5*1 genotypes and the CYP3A4*1B/CYP3A5*3 haplotype for associations with prostate cancer risk and tumor aggressivity (23). Consistent with findings from previous studies, those for comparison of CYP3A4*1B frequencies between prostate cancer cases and controls showed no simple associations (OR = 0.9, 95 percent CI: 0.6, 1.4 for Caucasians and OR = 1.0, 95 percent CI: 0.4, 2.5 for African Americans). However, when stratified by aggressivity, results indicated that CYP3A4*1B was a risk factor for more aggressive prostate cancer in Caucasians (OR = 1.9, 95 percent CI: 1.0, 3.6) but not African Americans (OR = 0.5, 95 percent CI: 0.2, 1.4). The CYP3A5*1 variant was inversely associated with prostate cancer in Caucasians with less aggressive disease (OR = 0.4, 95 percent CI: 0.2, 0.8) but not in African Americans (OR = 0.9, 95 percent CI: 0.1, 8.7). The CYP3A4*1B/CYP3A5*3 haplotype was positively associated with prostate cancer when data from African Americans and Caucasians were combined (OR = 2.9, 95 percent CI: 1.4, 6.2).

Overall, the data from the five prostate cancer studies described above do not provide convincing support for a direct role of CYP3A4*1B in prostate carcinogenesis (25, 26, 38, 60, 61). However, the apparent association of CYP3A4*1B with other factors such as age and clinical stages of disease will likely make this role difficult to define. In the studies from Scotland, CYP3A4*1B appears to play a role in high-risk (benign prostatic hyperplasia) patients. Two studies of Caucasians suggest an association with advanced disease only, whereas findings from studies of African Americans are inconsistent. A major gap is the dearth of robust, but basic case-control studies; future studies might consider gene-environment and gene-gene interactions in seeking an association between prostate cancer and CYP3A4 polymorphisms.

Although we know of no research that has been conducted to specifically evaluate interactions among CYP3A4 variants, breast cancer, and other external factors, CYP3A4’s role in the metabolism of exogenous chemicals makes it a likely candidate for gene-environment interactions. CYP3A4 is present in mammary epithelial cells (9, 10, 12, 13) and is involved in activation of many environmental carcinogens, such as the polycyclic aromatic hydrocarbons, heterocyclic amines, aflatoxin, and nitrosamines (14, 85-88). Some have been shown to be mammary carcinogens in laboratory animals (89-93). Furthermore, ingestion of polycyclic aromatic hydrocarbons and heterocyclic amines and metabolism via CYP3A4 has also been shown to result in formation of carcinogen-DNA adducts in mammary tissue (94). On the basis of this type of inferential evidence, the potential for a gene-environment interaction in the mechanism of breast cancer does seem biologically plausible.

The apparent complexity of the associations of CYP3A4*1B with prostate cancer risk in relation to age, clinical factors, and family history suggests that interactions with other probable endogenous factors are required. Thus, CYP3A4*1B is probably not an independent risk factor.

No standardized laboratory tests exist for CYP3A4 genotyping. A variety of methods for determining CYP3A4 genotype have been used and were referred to above, and they are listed in Table 8. These methods include dideoxy-chain termination DNA sequencing, real time-PCR, PCR-restriction fragment length polymorphism, single stranded conformation polymorphism, single nucleotide primer extension (SNuPe; Amersham Biosciences, Piscataway, New Jersey), and conformation sensitive gel electrophoresis.

Currently, there is insufficient evidence implicating DNA-sequence variation in CYP3A4 in the etiology of either prostate or breast cancer for population testing. Development of haplotyping methods that identify specific groups of these polymorphisms may help to resolve this question (95). However, considering the complexity of sex hormone metabolism, it is unlikely that a single nucleotide polymorphism in a single steroid metabolism gene will be sufficiently implicated to warrant genetic testing.

The authors gratefully acknowledge Glory Johnson for her invaluable editorial support.

Reprint requests to Dr. Ainsley Weston, National Institute for Occupational Safety and Health–CDC, 1095 Willowdale Road, MS-L3014, Morgantown, WV 26505-2888 (e-mail: agw8@cdc.gov).

• Table 1 - DNA sequence variants of CYP3A4 found in the enhancer region, promoter region, and 3'UTR
• Table 2 - DNA sequence variants of CYP3A4 found in the coding region
• Table 3 - DNA sequence variants of CYP3A4 found in introns
• Table 4 - Reported frequencies of CYP3A4*1B
• Table 5 - Reported frequencies of nonsynonymous polymorphisms of CYP3A4
• Table 6 - Allelic frequency* of CYP3A4 noncoding polymorphisms other than CYP3A4*1B
• Table 7 - Results of case-control and cohort studies of breast and prostate cancer
• Table 8 - Details concerning populations of studies cited and laboratory methods
• Figure 1 - The chemical structure of nifedipine
• Figure 2 - CYP3A4 and androgen metabolism
• Figure 3 - Incidence rates of breast and prostate cancer in five different ethnic groups, Surveillance, Epidemiology, and End Results Program, 1992–1998
• Figure 4 - Possible involvement of CYP3A4 in estrogen metabolism and breast cancer risk in Chinese women

List of References

• Human Cytochrome P-450 Allele Nomenclature (last accessed 4/2007)
• GeneCards (last accessed 4/2007)
• Cytochrome P-450 Home Page (last accessed 4/2007)
• GenAtlas   (last accessed 4/2007)
• Online Mendelian Inheritance in Man (OMIM) (last accessed 4/2007)
• Genome Database (GDB) (last accessed 4/2007)