  |
 |
 |
 |
 |
|||
 |
|||||||
 |
|
|||||||||||||||
|
Journal Publication |
|||||||||||||||||
A Review of Genetic Polymorphisms and Prostate Cancer Riskby Steven S. Coughlin, Ph.D.1 and Ingrid J. Hall Ph.D., M.P.H.11Epidemiology and Health Services Research Branch, Division of Cancer Prevention and Control, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA Reprint requests should be addressed
to: Family history is an established risk factor for prostate cancer and families demonstrating autosomal dominant or X-linked transmission of susceptibility have been observed (1-5). Although an increasing number of candidate genes for hereditary prostate cancer susceptibility have been identified (6-17), only 5 to 10 percent of prostate cancer cases in the population may arise from major susceptibility genes. Other genetic factors, in combination with possible environmental risk factors for prostate cancer, may have greater public health importance. Genetic polymorphisms that may be associated with prostate cancer risk are much more common in the population than are high-penetrance cancer susceptibility genes. The search for genetic markers for prostate cancer susceptibility has revealed an increasing number of relatively common genetic polymorphisms that may have a role in the metabolism of testosterone and other androgens (18-21). In the cell, more than 90% of free testosterone is irreversibly converted into the main prostatic androgen, dihydrotestosterone (DHT), through the enzymatic action of 5a-reductase (21, 22). DHT binds to the androgen receptor (AR) and forms a complex that, in conjunction with cofactor proteins and other transcription factors, facilitates androgen-induced regulation of genes involved in cellular proliferation and differentiation (22). When localized to the prostate, androgens are responsible for normal and hyperplastic growth (23). Suppression of DHT may reduce the carcinogenic transformation of prostate cells. Elevated levels of testosterone and intraprostatic DHT may partly account for racial differences in prostate cancer risk (24). In this paper, we review a variety of genetic polymorphisms that may have an etiologic role in prostate cancer. We include associations identified in molecular epidemiology studies and the consistency of findings reported to date. Suggestions for further research are also offered. For the purposes of this review, we identified relevant articles through a MEDLINE search for the period of January 1987 through March 2001. The searches were limited to articles published in English. Medical subject headings were used to scan titles, abstracts, and subject headings in the databases using the keywords prostate neoplasms,@ genetics, and polymorphisms.@ The search for genetic markers of disease susceptibility often utilizes the candidate gene approach, where a gene is targeted based on the properties and metabolic pathways of its protein product (25). For example, prostatic cell growth is dependent on testosterone, an androgen. Thus, genes involved in the androgen metabolism cascade have been identified as possible candidates for genetic influences in prostate cancer. When these genes are found to be polymorphic and the variants are distributed differently across population groups, interest in them increases as variation in DNA sequence could alter protein function and result in variations in disease risk. Analyses of this type are complicated by the fact that single nucleotide changes may be difficult to distinguish and all changes in DNA sequence do not result in differences in protein function (25). A futher complication is that disease may not be associated with the candidate gene but, rather, the candidate gene may be in linkage disequilibrium, or closely associated with a gene sequence in close proximity that is the actual causative agent. Despite widespread use, the candidate gene approach is limited by unidentified gene-disease associations and by uncharacterized protein function (25). Polymorphisms of Genes Involved with the Androgen-signaling Cascade Androgen Receptor Gene The AR gene, which is located on the X-chromosome (26), is polymorphic with a highly variable number of trinucletide microsatellite repeats of CAG and GGN in exon 1. The number of CAG repeats range between 8 and 31, with an average of about 20 (26). Longer AR variants have decreased transactivation activity and decreased binding affinity for androgens and may confer protection to the prostate (22, 26, 27). The relative increased responsiveness of short variants may elevate the risk of prostate cancer because of chronic androgen overstimulation (26). Some, but not all studies have shown that prostate cancer patients tend to have shorter CAG repeats than nonprostate cancer controls (28, 29), and African-Americans, who have generally shorter CAG repeat lengths in the AR gene (30), have a higher incidence and mortality rate from prostate cancer (26). Thus, functional differences among long and short AR variants may influence prostate cancer risk. Studies examining CAG and GGN repeat lengths in the AR gene have yielded conflicting results, as summarized in Table 1. Ingles et al. (29) examined associations with a polymorphic AR gene microsatellite (CAG repeat) in a case-control study of 57 non-Hispanic white men with prostate cancer and 169 non-Hispanic white neighborhood controls in Los Angeles County. The study was an extension of an earlier investigation by Irvine et al. (28). Men who carried AR CAG alleles with fewer than 20 repeats had a higher risk of prostate cancer compared with those who carried alleles with 20 or more repeats (adjusted OR = 2.1, 95% CI 1.1-4.0) (29). Moreover, short alleles were more strongly associated with advanced disease than with localized disease (OR = 2.4, 95% CI 1.0-5.5). In a case-control study of 587 newly diagnosed cases of prostate cancer and 588 controls nested within the Physician=s Health Study, Giovannuci et al. found that men with less than 18 CAG repeats had a higher risk of prostate cancer (RR = 1.5, 95% CI 0.9-2.5) than men with more than 26 repeats (19). Again, an association was observed for both advanced stage (RR = 2.2, 95% CI 1.1-4.6) and high grade (RR = 1.9, 95% CI 1.0-3.8) tumors. A related case-control study nested within the Physician=s Health Study found no significant differences in the mean GGN repeat length between incident cases of prostate cancer and control patients (31). However, the prostate cancer cases had a narrower spread of repeat lengths compared with controls. Stanford et al. (32) found that men with fewer CAG repeats (<22 vs. >22) had a slightly higher risk of prostate cancer in a population-based case-control study of prostate cancer among white men in Washington State (n = 257 cases and 250 controls, OR = 1.2, 95% CI 0.9-1.7). Results from this study also revealed an association between GGN repeat length and prostate cancer. Men with less than or equal to 16 GGN repeats had a higher risk of prostate cancer as compared with those with more than 16 repeats (OR = 1.6, 95% CI 1.1-2.4). Because of its population-based design, this study may offer enhanced scientific validity as compared with studies that have utilized convenience samples of cases and controls. An association with short CAG repeat length was also observed by Hsing et al. (33) in a recent population-based case-control study in China (190 incident cases and 304 age-matched controls). The median repeat length among this population was longer--23 repeats--than that observed for Western men. In the case-control comparisons, men with less than 23 CAG repeats had an increased risk of prostate cancer (OR = 1.7, 95% CI 1.1-2.4), but no significant association was observed for GGN repeat length greater or less than 23 repeats. These findings are inconsistent with those arising from other studies of AR CAG or GGN repeat length (34-38)(Table 1). For example, Bratt et al. (34) did not observe an association with CAG repeat length in a clinical study of prostate cancer in Sweden. Nor did Ekman et al. (35), who conducted an international comparative study of AR CAG repeats among 59 Swedish patients with sporadic prostate cancer, 59 with hereditary prostate cancer, 34 Japanese prostate cancer patients, and 69 controls with benign prostatic hypertrophy. Although the CAG repeats were shorter among Swedish prostate cancer cases than among Swedish controls, the Japanese prostate cancer patients had longer repeats than did Japanese controls. No difference was observed in number of CAG repeats between sporadic and hereditary prostate cancer patients. Likewise, Lange et al. (36) found no association with AR CAG repeat length among 270 white prostate cancer patients and 305 controls, and Edwards et al. (37) found no association with either CAG or GGN repeat length in DNA samples from 178 British prostate cancer cases and 295 age- and location-matched controls who were women. Correa-Cerro et al. (38) examined these repeats in a case-control study of French and German prostate cancer patients. The study included 132 cases of sporadic prostate cancer, 105 convenience controls, and a sample of families with prostate cancer (85 affected and 46 unaffected family members). No statistically significant associations with CAG or GGC repeats were observed in case-control comparisons or in linkage analysis. Mononen et al. (39) examined associations with a genetic variant at codon 726 in exon E of the AR gene (the R726L variant, an arginine to leucine change) among 418 prostate cancer patients in Finland who had no reported family history of the disease and 106 prostate cancer patients who had a positive family history. The population frequency of the R726L variant in blood donors was 3 of 900 (0.33%). In contrast, eight (1.9%, OR = 5.8, p = 0.006) were found in the group of patients with sporadic prostate cancer, and two (1.9%) among patients with a positive family history (OR = 5.8, p = .09). Although these interesting findings suggest avenues for future research, the study is not population-based and is also limited by the lack of a rigorous control group and control for possible confounding factors. 5a-Reductase Type II Gene Two isozymes of 5a-reductase exist. The gene encoding 5a-reductase type I is located on chromosome 5, and the protein is expressed in the liver, skin, and scalp (24). The gene that codes for 5a-reductase type II (SRD5A2) is located on chromosome 2. It is expressed in the prostate and in the androgen-sensitive cells of genital skin where testosterone is converted irreversibly to DHT by 5a-reductase type II (24). Evidence suggests that 5a-reductase type II activity is reduced in populations at lower risk for prostate cancer, including Chinese and Japanese men (23, 40). A polymorphism in the UTR of the SRD5A2 gene has been identified (41). Ten alleles fall into three families that differ in the number of TA dinucleotide repeats; some alleles are found exclusively among Asians and others among African-Americans (24, 42). Although no clinical significance for these polymorphisms has yet been determined, some TA repeat alleles may promote an elevation of enzyme activity, which may in turn increase the level of DHT in the prostate (21, 22, 24). Thus, these polymorphisms may increase the risk of prostate cancer. Reichardt et al. (24) found an excess of longer alleles among 59 white prostate cancer cases compared to an equal number of age-matched controls (Table 2). However, Kantoff et al. (42), who examined associations with TA dinucleotide repeat polymorphisms in a much larger case-control study nested within the Physician=s Health Study, found conflicting results. A total of 590 predominately white men with incident prostate cancer and 802 age-matched controls were included in this study. Homozygosity for the longer TA allelles, TA(9) and TA(18), was not associated with higher prostate cancer risk. Rather, a possible inverse association of borderline significance (OR = 0.5, 95% CI 0.2-1.1) was found. Longer TA alleles were found to be more prevalent among men without prostate cancer (42). Although not population-based, the study offers the advantages of a large sample size and adequate control for confounding factors. In a population-based study in Los Angeles and Hawaii, Makridakis et al. (43) reported that a second SRD5A2 polymorphism, which replaces valine with leucine at codon 89 (V89L), is correlated with decreased 5a-reductase type II activity and is differentially distributed among racial and ethnic groups. The V89L substitution resulted in an almost 30% reduction in enzyme activity and was most common among Asian Americans. African Americans were found to have the lowest frequency for the V89L amino acid substitution. Using placental DNA, Lunn et al. (44) confirmed the higher frequency of the V89L allele among Taiwanese. Among 108 prostatectomy patients (96 white and 12 African American) at one hospital in North Carolina and 167 male non-cancer urology clinic controls, African-American men had the lowest prevalence of the V89L substitution. However, no protective effects of the leucine allele were observed. The V89L allele was somewhat more common among prostatectomy patients than controls (OR = 1.3, 95% CI 0.8-2.2), but this difference was not statistically significant. Febbo et al. (45) also examined the V89L polymorphism, in a nested case-control study conducted within the Physicians= Health Study. No difference was observed in the distribution of genotypes of the V89L polymorphism among 584 incident cases of prostate cancer and 799 matched controls. Compared with individuals who were homozygous for the val allele, those with 1 or 2 leucine alleles (val/leu or leu/leu) were not at significantly reduced risk (OR = 1.0, 95% CI 0.8-1.2 and OR = 0.8, 95% CI 0.6-1.2, respectively). Makridakis et al. (20) reported a third variant of the SRD5A2 gene in which an alanine residue at codon 49 is replaced with threonine (A49T). A nested case-control design was used to assess the risk of developing prostate cancer among 388 cases (216 African-American, 172 Hispanic) and 461 controls (261 African-American, 200 Hispanic) selected from a population-based cohort study in Los Angeles and Hawaii. The variant allele was rare among the controls, but it conferred a relative risk for both advanced and localized cases of prostate cancer of 3.3 (95% CI 1.1-11.9) among African-American men and 2.5 (95% CI 0.9-7.4) among Hispanic men. Risk of advanced prostate cancer was increased among African-American (RR = 7.2, 95% CI 2.2-27.9) and Hispanic men (RR = 3.6, 95% CI 1.1-12.3). In a separate study that did not include a control group, the A489T variant was observed among Caucasians and was found to be associated with pathological characteristics of prostate cancer (higher frequency of extracapsular disease and greater lymph node metastasis) (46). 3b - Hydroxysteroid Dehydrogenase Type II Gene The 3-hydroxysteroid dehydrogenases regulate DHT levels by inactivating DHT which is ultimately excreted (18). The type II isozyme form of 3b-hydroxysteroid dehydrogenase, which is encoded by the HSD3B2 gene, is primarily expressed in the adrenal glands, testis, ovary, and prostate. Devgan et al. (18) examined the distribution of a dinucleotide repeat in the HSD3B2 gene in 312 men across three racial groups: African Americans, Asians, and white men. The 289 bp allele was the most common allele in all population groups, but was more common among white men (allele frequency 51.6%), compared with African Americans (allele frequency 33.6%) or Asians (allele frequency 36.7%). Alleles ranging from 302 to 334 bp and a 281 bp allele were more common among Asians and may be protective. However, no studies have elucidated disease risk associated with a particular allele. Prostate Specific Antigen Gene The DHT-AR complex regulates gene transcription by binding DNA sequences known as androgen responsive elements (AREs) in the promoters of target genes. Although the specific target genes are not known, one candidate, the PSA gene, is androgen-regulated in the prostate. Rao and Cramer (47) found that the ARE sequence of the PSA gene promoter was polymorphic. Xue et al. (48) examined the prostate cancer risk associated with the G allele, a G to A substitution at position B158 in the promoter region. Homozygosity for the G allele yielded an odds ratio of 1.6 (95% CI 0.7-3.8) when compared with homozygosity for the A allele among non-Hispanic white men (57 cases and 156 controls) in Los Angeles County. A stronger association was observed for advanced cases (OR = 2.9, 95% CI 1.2-6.8). No increase in risk was observed for heterozygotes (OR = 1.1, 95% CI 0.5-2.3). Yang et al. (49) examined two novel polymorphisms in the ARE region of the PSA gene promoter among 47 Japanese prostate cancer cases and 105 controls. Two alleles (A-AA and G-A) were present. The proportion of cases and controls who were either heterozygotes or homozygotes for the A-AA allele was not significantly different (p = .726). However, in comparison to G-A allele homozygotes, prostate cancer patients with at least one A-AA allele tended to have higher serum PSA levels (p = .0002), poorer differentiation (p = .015) and more advanced clinical stage (p = .008). An increasing number of studies have examined whether polymorphisms in the vitamin D receptor (VDR) gene are related to prostate cancer risk. Several lines of evidence suggest that vitamin D may influence prostate cancer risk. For example, the increased incidence of prostate cancer among individuals living at northern latitudes suggests that lack of exposure to ultraviolet radiation is a possible risk factor for the disease (50). The active hormonal form of vitamin D, 1,25-dihydroxyvitamin D (1,25-D), influences prostate cell division through the VDR (28, 51). In prostate cancer cell lines, 1,25-D has been shown to induce cell differentiation and inhibit proliferation (52), and in some epidemiologic studies, a relationship has been observed between higher serum levels of 1,25-D and lower risk of prostate cancer, but results from reported studies have been inconsistent (53-55). A poly-A microsatellite repeat in the 3' UTR of the VDR gene has been identified in studies among white men. However, this polymorphism is in strong linkage disequilibrium with several restriction fragment length polymorphisms (RFLPs) located in intron 8 (BsmI and ApaI) and exon 9 (TaqI) (28, 50, 65). Neither the TaqI nor the BsmI restriction site have been shown to directly affect VDR function (56). Although the 3= UTR region is involved in regulating mRNA stability (57), the poly-A microsatellite polymorphism has not been shown to have functional consequences (58). All of these polymorphisms may identify the same genotype, and they may be in linkage disequilibrium with other variants that alter VDR function. Molecular epidemiology studies of VDR polymorphisms and prostate cancer are summarized in Table 3. Taylor et al. (56) found an overall odds ratio of prostate cancer of 0.3 (95% CI 0.2-0.8) for homozygosity for the TaqI-t allele, as compared with TT/Tt genotype, in a case-control study of 108 men undergoing radical prostatectomy and 170 male urology clinic controls with no history of cancer. Other studies of prostate cancer, however, have failed to confirm an association with homozygosity for the TaqI-t allele (50, 52). In a case-control study of 372 incident cases and 591 controls nested within the Physicians= Health Study, Ma et al. (52) found an overall odds ratio of developing prostate cancer of 0.9 (95% CI 0.6-1.4) for homozygosity for the TaqI-t allele, as compared with TT. They did not observe significant associations even after they stratified by age or tumor aggressiveness (52). Kibel et al. (50) also failed to observe a relationship between fatal prostate cancer and homozygosity for the TaqI-t allele. In a case-control study of 41 patients who died of prostate cancer and 41 controls with no evidence of prostate cancer, Kibel obtained an overall odds ratio of metastatic prostate cancer of 0.7 (0.2-2.6) for homozygosity for the TaqI-t allele compared with TT and Tt combined (50). In a fourth study of 66 Japanese prostate cancer patients and 60 noncancer urology clinic controls (59), homozygosity for the TaqI-t allele was rare. Homozygosity or heterozygosity for the TaqI-t allele was associated with an odds ratio of 1.3 (95% CI 0.6-2.8, tt/Tt vs. TT); there were too few cases homozygous for the TaqI-t allele to estimate the odds ratio for the comparison of tt vs TT/Tt. Blazer et al. (60) also found a statistically nonsignificant association with homozygosity for the TaqI-t allele in a case-control study in North Carolina (77 cases and 183 controls). In this study, the adjusted OR for the comparison of TT/Tt vs tt was 1.4 (95% CI 0.7-2.9). To allow comparisons across studies, we estimated the odds ratio for the comparison of tt vs TT/Tt to be about 0.7 by taking the reciprocal of the adjusted odds ratio. Little evidence supports an association between the BsmI polymorphism and prostate cancer risk. Ma et al. (52) found an overall odds ratio of 0.9 (95% CI 0.6-1.3) for homozygosity for the BsmI-B allele. To allow comparisons across studies, we estimated the odds ratio for the comparison of bb vs BB/Bb to be about 1.1. Although they observed no significant associations with homozygosity for the BsmI-B allele after stratification by age or tumor aggressiveness, these investigators observed an odds ratio of developing prostate cancer of 0.4 (95% CI 0.2-1.0) for homozygosity for the BsmI-B allele among men with plasma 1, 25-D levels below the median (52). In contrast, Ingles et al. (58) found BsmI-bb alleles to be associated with protection against advanced disease in 151 African-American incident cases and 174 controls (OR = 0.4, 95% CI 0.2-1.0) although this effect was not observed among all cases (OR = 1.1, 95% CI, 0.5-2.4). A third study (61) examined the effect of the BsmI polymorphism on prostate cancer risk among 222 Japanese prostate cancer patients and 128 male noncancer controls. A strong association with prostate cancer was found for homozygosity for the BsmI-B allele (age-adjusted OR = 3.3, 95% CI 2.1-5.3), compared with BB and Bb combined. Ingles et al. (29) found an overall odds ratio of 4.5 (95% CI 1.3-16.1) for the third VDR polymorphism, the poly-A long allele, in a case-control study involving 57 non-Hispanic white men with prostate cancer and 169 non-Hispanic white neighborhood controls. Other studies of prostate cancer, however, have found no association with the poly-A microsatellite repeat polymorphism (58). Blazer et al. (60) did not detect an association with long poly-A alleles among 77 cases and 183 controls in a North Carolina study. The odds ratio for the long poly-A alleles (LL/SL versus SS) was 1.2 (95% CI 0.6-2.4). Kibel et al. (50) also found no association with poly-A alleles in a study of 41 lethal metastatic prostate cancer cases and 41 noncancer urology patients. The odds ratio for the comparison of SS genotype with LL and SL combined was 0.8 (95% CI not provided). For Kibel=s study, we estimated the odds ratio for the comparison of LL and SL genotypes vs. SS to be 1.3 (95% CI 0.4-4.3). Interestingly, results from this study showed that the poly-A microsatellite is in strong linkage disequilibrium with the Taq1 restriction fragment. These results suggest that Taylor et al. (56) and Ingles et al. (29) may have been examining the same VDR genotypes. In contrast, Correa-Cerro et al. (62) detected significant protective associations only with the heterozygous genotypes of poly-A (SL vs SS, OR = 0.4, 95% CI 0.2-1.0) and TaqI-tt (Tt vs TT, OR = 0.5, 95% CI 0.3-0.9). Phase I enzymes of the cytochrome P450 (CYP) family activate chemical carcinogens to electrophilic reactive forms to produce DNA adducts (63). Cytochrome P450 enzymes make up a multiple-gene superfamily that plays an important role in steroidogenesis and in the activation or detoxification of environmental chemicals such as polycyclic aromatic hydrocarbons, benzo(a)pyrene, and heterocyclic amines (64, 65). In contrast, phase II enzymes, such as the glutathione S-transferase (GST) family, facilitate excretion of carcinogenic compounds by conjugating metabolic intermediates to water-soluble forms. Therefore, persons with increased metabolic activity and decreased detoxifying activity of phase l and ll enzymes may be at higher risk of prostate cancer (66). Molecular epidemiology studies of prostate cancer have examined associations with phase I and phase II genes, such as CYP1A1, CYP2D6, CYP17A2, CYP3A4, GST, and NAT1 and NAT2. Cytochrome P450 Enzymes The CYP1A1 gene encodes aryl hydrocarbon hydroxylase (AHH), which is primarily expressed in liver but has been detected in prostate tissue (67). The highly inducible AHH catalyzes the monooxygenation of polycyclic aromatic hydrocarbons to phenolic products and epoxides, both of which may be carcinogenic. Although several restriction fragment length polymorphism (RFLP) patterns in the CYP1A1 gene have been identified, three have received the most attention, including a MspI RFLP (CYP1A1*2A) at the 3' end of the gene (68). The MspI RFLP (CYP1A1*2C) is closely linked to another polymorphism in which an A to G point mutation in exon 7 replaces isoleucine 462 with valine. This allele produces an enzyme with higher catalytic activity than the Ile/Ile homozygote (69). Higher catalytic activity may predispose patients to prostate cancer by increasing the amount of carcinogenic epoxides and phenolic products in prostate tissues. A third polymorphism (CYP1A1*3), also in the 3' end of the gene, has been identified among African-Americans (AA) (68). A fourth substitution (CYP1A1*4) has been described in exon 7 that exchanges threonine to asparagine at codon 461. Murata et al. (63) examined associations between prostate cancer and the exon 7 polymorphism of the CYP1A1 gene. The study involved 115 prostate cancer patients and 204 control patients in Japan. Both homozygosity for the valine allele (Val/Val) (OR = 2.6, 95% CI = 1.1-6.3) and heterozygosity for the valine allele (Ile/Val) (OR = 1.4, 95% CI = 0.9-2.3) were associated with increased risk of prostate cancer, compared with homozygosity for the isoleucine allele (Ile/Ile). The CYP2D6 gene encodes for debrisoquine hydroxylase, which metabolizes a variety of drugs and other xenobiotics (64, 65). Debrisoquine hydroxylase may activate procarcinogens or, conversely, detoxify carcinogens. A number of alleles at the CYP2D6 locus produce enzymes that are unable to efficiently metabolize debrisoquine and similar compounds. Persons homozygous for a characteristic G to A transition variant (CYP2D6*1B) at the junction of exon 3 and intron 4, (or other variant alleles that are less common) have the poor metabolizer phenotype and are very sensitive to the drugs hydroxylated by this enzyme. Molecular epidemiology studies that have examined associations of prostate cancer with CYP2D6 polymorphisms have yielded conflicting results (Table 4). Febbo et al. (70) found an overall odds ratio of 1.4 (95% CI 0.9-2.2) for homozygosity for the CYP2D6-B allele in a case-control study of 571 incident cases and 767 controls nested within the Physicians= Health Study. The relative risk for aggressive prostate cancer was similar to the overall risk for prostate cancer among patients who were homozygous for the poor metabolizer allele (OR = 1.3, 95% CI 0.7-2.4). In addition, a small increase in the risk of prostate cancer was observed among men with one copy of the B allele (OR = 1.2, 95% CI 0.9-1.5). No evidence of an interactive effect with cigarette smoking was observed (70). Wolf et al. (71) also found a possible increased risk associated with homozygosity for the CYP2D6-B allele or B/deletion alleles (odds ratio = 2.4, 95% 0.8-7.1) in a comparison of 54 prevalent cases of prostate cancer and 720 clinic and volunteer controls, but here the association was not statistically significant. Nonsignificant results were also obtained by Agundez et al. (67) in a case-control study in Spain (94 prostate cancer patients, 160 volunteer controls); the OR for the poor metabolizer phenotype was 1.4 (95% CI 0.4-4.6). In a case-control study of 178 Swedish cases and 160 controls combined with 153 Danish cases and 359 controls (72), poor metabolizers, here defined by phenotype rather than CYP2D6 genotype, had an increased risk of prostate cancer (OR=1.4, 95% CI 0.8-2.4). Risk was increased for all smokers (OR = 2.1, 95% CI 1.0-4.3) and for Danish smokers (OR = 3.1, 95% CI 1.1-8.9). However, the significance of the apparent interaction with smoking observed in this study (72) is unclear, because no definite carcinogen for prostatic epithelium has been identified. In addition, no direct evidence supports androgens as substrates for debrisoquine hydroxylase. The B allele may be in linkage disequilibrium with an unidentified gene that alters the risk of prostate cancer (70). Polymorphisms have been identified in the CYP17 gene which is involved in androgen biosynthesis and metabolism. The A2 allele contains a T to C transition that creates an additional Sp1-type promoter site in the 5' promoter region. Lunn et al. (44) found that the A2 allele (combined A1/A2 and A2/A2 genotypes) occurred at a higher frequency in 96 white patients with prostate cancer (70%) than in 159 white urology clinic controls (57%). This result suggests that the A2 allele may convey increased risk for prostate cancer (OR = 1.7, 95% CI 1.0-3.0). Small sample sizes limited analyses for African Americans alone, but results were the same when all study participants were pooled. Among those who were less than 64 years of age at diagnosis, the presence of A2 alleles was significantly associated with prostate cancer (OR = 2.3, 95% CI 1.0-5.4). Gsur et al. (73) also found an association of the CYP17 A2 allele with prostate cancer. In a case-control study (63 cases and 126 BPH controls) from Austria, homozygosity for the A2 allele conferred an increased risk (OR=2.8, 95% CI 1.0-7.8). The odds ratio was 8.9 (95% CI 1.8-49.2) among patients over 66 years of age. The A1 allele was also found to be associated with prostate cancer in a Swedish study conducted by Wadelius et al. (74). Among 178 prostate cancer patients and 160 age-matched population controls, an odds ratio of 1.6 (95% CI 1.0-2.5) was observed for homozygosity for the A1 allele. In a case-control study (252 cases and 131 controls) from Japan (75), men who were homozygous for the A1 allele had an increased risk of prostate cancer (OR=2.6, 95% CI 1.4-4.8) compared with those homozygous for A2. Men who were heterozygous had an intermediate increased risk of prostate cancer (OR=1.5, 95% CI 0.8-2.5). There was no significant association between CYP17 polymorphisms and tumor grade or stage. CYP3A4 is a gene involved in the oxidation of testosterone. Rebbeck et al. (76) identified a genetic variant of CYP3A4 (CYP3A4*1B) that was associated with a higher clinical grade and stage in white men with prostate cancer. The allele frequency of this variant, G, is differentially distributed across racial and ethnic groups (77). In a convenience sample of 117 white, 121 Hispanic, 116 African American, and 80 Asian American healthy volunteers, Paris et al. (78) found that the frequency of the variant G allele was much more common among African American men (81%) than white (7%), Hispanic (20%), or Asian American (0%) men. Among 174 African-American prostate cancer cases, 83% carried at least one variant allele; 28% of 116 African-American controls were homozygous for the G allele. This frequency increased to 46% among African-American prostate cancer cases, and homozygosity for the G allele correlated with clinical characteristics (Gleason score >7, PSA>10, and higher grade/stage). The variant m1 allele (CYP2C19*2A) of the CYP2C19 gene was also evaluated for its association with prostate cancer risk among a Swedish population (72). No significant differences were found between cancer patients and controls (OR = 1.1, 95% CI 0.3-4.3). N-Acetyltransferase 1 and 2 The N-acetyltransferase genes, NAT1 and NAT2, are polymorphically expressed in a variety of tissues; NAT2 is expressed in the prostate (67, 79). Both isozymes catalyze the metabolic activation of aromatic amines and heterocyclic amine carcinogens such as 4-aminobiphenyl which is found in tobacco smoke (64). The enzymes also deactivate, or detoxify, these compounds. Thus, genetic polymorphisms in NAT1 and/or NAT2 may modify the cancer risk related to exposures to these carcinogens (64, 65). Numerous genetic variants of NAT1 and NAT2, which result in alterations of acetylator phenotype, have been identified (79). Racial and ethnic differences in NAT1 and NAT2 genotype frequencies may be a factor in cancer incidence (79). Agundez et al. (67) identified 20 different NAT2 genotypes and evaluated their associations with risk in a study of 94 cases of prostate cancer and 160 volunteer controls. No statistically significant differences in the prevalence of NAT2 gene variants were observed between cases and controls (67). The odds ratio for the slow acetylation phenotype, which was predicted by multiple NAT2 genotypes, was 1.1 (95% CI 0.7-1.9). Nor was any association observed in a case-control study of Swedish and Danish persons (338 cases and 512 controls) (72). In this study, the odds ratio for risk of prostate cancer among slow acetylators, which were identified by NAT2 genotype, was 0.97 (95% CI 0.7-1.3). Fukutome et al. (80) examined associations between NAT1 polymorphisms and prostate cancer in a study of 101 patients and 97 controls. Homozygosity for the NAT1*10 allele, a variant associated with the rapid acetylator phenotype, was associated with higher risk (OR = 2.4, 95% CI 1.0-5.6), compared with a single NAT1*10 allelle or no NAT1*10 alleles. In addition, the frequency of NAT1*10 homozygous alleles was significantly higher among stage D (OR = 2.6, 95% CI 1.0-7.2), well-differentiated (OR = 3.3, 95% CI 1.2-9.5), and poorly differentiated adenocarcinomas (OR = 3.8, 95% CI 1.0-5.8). Glutathione S-Transferase Glutathione S-transferases (GSTs) catalyze the conjugation of glutathione to numerous potentially genotoxic compounds, including aliphatic aromatic heterocyclic radicals, epoxides, and arene oxides (64, 65, 81). Four families of enzymes have been classified as a, m, p, and θ. Glutathione S-transferase-mu (GSTM1), glutathione S-transferase-theta (GSTT1), and glutathione S-transferase-pi (GSTP1) have been studied most. GSTM1 is polymorphically expressed, and three alleles at the GSTM1 locus have been identified: GSTM1-0, GSTM1a, and GSTM1b (81). Two functionally different genotypes at the GSTT1 locus have been described: GSTT1-0, a homozygous deletion genotype, and GSTT1-, which is genotypes with one or two undeleted alleles. There are also three variant alleles of GSTP1. GSTP1a and GSTP1b differ by an A to G substitution at codon 105. GSTP1c varies from the a allele by the substitution of valine for alanine at codon 114 (72). Individuals with homozygous deletions of GSTM or GSTT lack glutathione S-transferase and therefore may be unable to eliminate electrophilic carcinogens as efficiently, which may increase the risk of somatic mutations leading to tumor formation. Alternatively, GSTT1 is expressed at high levels in the prostate, which suggests that the activation of compounds to genotoxic intermediates by GSTT1 could increase prostate cancer risk (82). Murata et al. (63) examined associations between prostate cancer and GSTM1 polymorphisms in a study of 115 Japanese prostate cancer patients and 204 controls, as shown in Table 5. Homozygosity for the GSTM1-0 allele was not significantly associated with increased risk overall (OR = 1.3, 95% CI = 0.8-2.0). Greater ORs with advancing stage were noted but the differences were not significant. The OR was 0.8 (95% CI = 0.4-1.6) for stages A and B, 1.3 (95% CI 0.6-3.0) for stage C, and 1.7 (95% CI 0.9-3.2) for stage D. Rebbeck et al. (82) evaluated GSTM1 or GSTT1 deletion polymorphisms in a study of 237 incident prostate cancer cases and 239 clinic controls from the U.S. Men who had nondeleted (functional) alleles at GSTT1 had a higher risk of prostate cancer (OR = 1.8, 95% CI 1.2-2.8), but this was not true of those who had nondeleted alleles at GSTM1 (OR = 1.1, 95% CI 0.7-1.6). No interactive effects between GSTM1 or GSTT1 were observed. Autrup et al. (83) examined associations with GSTM1, GSTT1, and GSTP1 polymorphisms in a case-control study of 153 prostate cancer cases and 288 controls in Denmark. Homozygosity for the GSTM1-0 allele was not associated with prostate cancer (OR = 1.3, 95% CI 0.9-1.9). Neither homozygosity for the GSTT1-0 allele (OR = 1.3, 95% CI 0.8-2.2) nor homozygosity or heterozygosity (ab or bb) for the GSTP1b allele (OR = 0.8, 95% CI 0.5-1.2) were associated with prostate cancer risk. However, the association with deletion of both GSTM1 and GSTT1 was of borderline significance (OR = 1.7, 95% CI 0.9-3.4). Associations with GSTP1 polymorphisms were examined by Harries et al. (84) in a case-control study of 36 prostate cancer cases and 155 volunteer controls in Great Britain. GSTP1b homozygosity was not significantly associated with reduced risk of prostate cancer (OR = 0.4, 95% CI 0.02-3.3, p=.693). Wadelius et al. (72) examined associations with GSTP1 alleles, a, b, and c in a study of 171 Swedish prostate cancer cases and 148 controls. The allelic variants were equally distributed among the cases and controls and no significant associations were observed (72). Few studies have investigated the combined effect of polymorphic alleles in different genes on prostate cancer risk and not all studies have had a sufficient sample size or statistical power to adequately examine gene-gene interactions. Murata et al. (63) analyzed the combined effect of the CYP1A1 and GSTM1 polymorphisms. The combination of the CYP1A1 valine allele (Ile/Val and Val/Val) with GSTM1 (0/0) resulted in an OR of 2.3 (95%CI 1.2-4.5), compared with the Ile/Ile genotype combined with the GSTM1 positive genotype. This risk was greater than the risk associated with either allele alone. These results suggest that elevated metabolic activation and lowered detoxification of endogenous or exogenous carcinogens increase DNA adduct formation and therefore influence prostate cancer risk (63). Rebbeck et al. (82) evaluated the combined effect of polymorphisms in the GSTM1 and GSTT1 genes. Persons who carried null genotypes at both loci were not at increased risk (OR = 1.2, 95% CI 0.6-2.4), compared with those who carried GSTM1 positive and GSTT1 null genotypes. However, persons positive at both loci (OR = 2.0, 95% CI 1.1-3.7) or positive for GSTT1 only (OR = 1.9, 95% CI 1.0-3.4) were at increased risk. Thus, risk was increased only among those with a functional GSTT1 allele. In contrast, Autrup et al. (83) reported the highest risk among those null for both genes (OR = 1.7, 95% CI 0.9-3.4). Autrup et al. also examined disease risk associated with GSTM1/GSTP1 and GSTT1/GSTP1 genotypes and observed no significant associations. Stanford et al. (32) examined the combined effects of AR CAG and GGN repeats and found that long AR CAG alleles (>22 repeats) and short GGN alleles (<16 repeats) did not increase risk (OR = 1.2, 95% CI 0.6-2.4). However, short CAG alleles together with long GGN alleles modestly increased risk (OR = 1.5, 95% CI 0.8-2.9), and the combination of short CAG repeats and short GGN repeats significantly increased risk (OR = 2.1, 95% CI 1.1-3.8). Hsing et al. (33) found a similar result when they compared combined genotypes to the referent of long CAG repeats (>23) and long GGN repeats (>23). They found a possible increased risk for long CAG alleles and short GGN alleles (OR = 1.5, 95% CI 0.8-2.9), an increased risk for the combination of short CAG repeats (<23) and long GGN repeats (>23) (OR = 1.9, 95% CI 1.2-2.8), and a possible increased risk (OR = 1.8, 95% CI 0.9-3.4) for the combination of short CAG repeats with short GGN repeats. Xue et al. (48) found that the combined effect of two high-risk polymorphisms in the AR CAG repeat and in PSA conferred a higher risk for prostate cancer than did either polymorphism alone. Compared with those carrying the referent genotype, PSA AA (or AG) and AR CAG-long (>20 repeats) alleles, men who were homozygous for the PSA G allele with an AR CAG-long allele were at increased risk of advanced disease (OR = 4.2, 95% CI 1.3-13.6). Men who were not homozygous for the PSA G allele but who did have a short AR allele were at increased risk of advanced disease (OR = 3.6, 95% CI 1.1-11.6). Men who were homozygous for the PSA G allele and had short AR CAG repeats were at increased risk of prostate cancer (OR = 5.1, 95%CI 1.6-16.2) and of advanced disease (OR = 10.7, 95% CI 2.3-50.5). The etiology of prostate cancer is unlikely to be explained by allelic variability at a single locus. Instead, the prevalence of prostate cancer in the population probably results from complex interactions among many genetic and environmental factors over time (22, 81, 85, 86). The majority of prostate cancer cases are unlikely to be due to major susceptibility genes (6-17) and genetic polymorphisms are likely to be more important from a public health perspective. Cumulative lifetime exposure to androgens, androgen metabolites, and other physiologic factors, as well as environmental exposures, could play an important role in the development of prostate cancer in genetically predisposed men. An improved understanding of the interplay of xenobiotic exposures, endogenous physiology, and genetic variability at multiple loci may help to identify men who are at increased risk for prostate cancer. For many of the genetic polymorphisms discussed in this review, associations with prostate cancer have been inconsistent across studies, which may be due to methodologic limitations of some studies (for example, the use of prevalent rather than incident cases), sample size limitations, or differences in the underlying frequencies of alleles. Many of the studies identified in this review used convenience samples of cases and controls (for example, clinic patients, family members, or blood donors) and were not truly population-based. In some studies, there may have been inadequate control of confounding factors. The prevalence of purported environmental risk factors and the prevalence of unidentified etiologic factors may also vary across populations. The causality of associations between genetic polymorphisms and prostate cancer is uncertain because of the inconsistency across studies, but the associations are biologically plausible. The specificity of the associations is a causal criterion unlikely to be met since genetic polymorphisms may influence the risk of a variety of diseases. This review considered molecular epidemiology studies of genetic polymorphisms and prostate cancer risk, including evidence for gene-gene interactions and results from studies that have included cases and controls from different racial and ethnic groups. Racial differences in genetic polymorphisms that have a role in the metabolism of testosterone and other androgens may partly account for racial differences in prostate cancer risk (24). However, social factors, environmental factors, and gene-environment interactions may also account for differences in prostate cancer incidence and mortality (26, 35). Several authors have noted that, in general, genetic variation within racial and ethnic groups is greater than the genetic variation that exists across racial, ethnic, or cultural groups (87, 88). The incidence of prostate cancer and mortality from the disease are almost two-fold higher among African American men than among white men (86). Marked international differences in prostate cancer incidence and mortality have also been reported. For example, incidence and mortality rates among Chinese and Japanese men are relatively low (86). These differences may arise from genetic, environmental, or social influences. Further population-based studies of incident cases of prostate cancer, with adequate sample sizes in racially diverse populations, are needed. Large-scale studies that test for multiple genetic markers are likely to find some associations with prostate cancer to be statistically significant, due to multiple comparisons and because genetic markers may be in linkage disequilibrium. Understanding findings from such studies will require a careful consideration of biologic plausibility and an understanding of biological pathways. Future studies that examine associations among several genetic polymorphisms should take into account risk factors for prostate cancer, such as diet and other environmental exposures, as well as possible biological pathways. Of particular interest are studies of gene-environment interactions and gene-gene-environment interactions (89). To date, molecular epidemiology studies of prostate cancer have rarely looked at a variety of potential gene-environment interactions or explored associations and interactions with more than one genetic polymorphism. Few of the studies identified in this review attempted to account for why some prostatic cancers remain latent and others are clinically detected. Most men with prostate cancer die with the disease rather than from it (22). However, the initiation and progression of prostate cancer most likely result from a complex series of genetic events and environmental influences (86), and further studies are needed to identify potential causes for the distinction between latent and symptomatic prostatic cancer.
|
|||||||||||||||||
