Red Wine Consumption is Inversely Associated with 2-Amino-1-Methyl-6-Phenylimidazo[4,5-b]Pyridine–DNA Adduct Levels in Prostate

Prostate cancer occurs more often and with a greater loss of life in men of African descent in the United States than in whites. The most recent figures show a 60% higher prostate cancer incidence rate in African American men and 140% higher mortality rate (1). In the Cancer Prevention Study II Nutrition Cohort, total red meat intake was associated with higher risk of prostate cancer only in African American men and was mainly because of higher consumption levels of cooked processed meats (2). Heterocyclic amines are present in higher amounts in fried or grilled meats with the predominant heterocyclic amine compound being amino-1-methyl-6-phenylimidazo [4,5-b] pyridine (PhIP). PhIP has been shown to be carcinogenic in rat prostate (3, 4) and may derive its carcinogenic potential through the formation of DNA adducts. On the basis of dietary survey (5) and urinary excretion data (6), PhIP intake is higher in African Americans than whites. PhIP dietary consumption has also been shown to be associated with higher PSA levels in healthy African American men (7), and we have previously reported a positive association between PSA level and PhIP–DNA adducts in African Americans that was not observed in whites (8).

Beverages such as coffee, beer, and wine may diminish PhIP mutagenicity by altering the PhIP metabolism pathway (9–11). Survey data suggest that whites consume more of these three beverages compared with African Americans (12). The antioxidant polyphenol, resveratrol, found in red wines, has also been shown to inhibit PhIP–DNA adduct formation presumably by O-acetyltransferase and sulfotransferase catalysis (11). In rodents, beer inhibits PhIP-induced tumorigenicity and DNA adduct formation most likely by inhibiting enzymes involved in PhIP metabolism (9).

In addition to racial differences in PhIP metabolism, inherited genetic differences in enzymes that metabolize PhIP may also contribute to variation in the biological effects of this compound across populations. Among the nine sulfotransferases known to be expressed in human tissues, SULT1A1, 1A2, and 1A3 catalyze the sulfate conjugation of many phenolic compounds and other xenobiotics, such as PhIP (13). A common single nucleotide polymorphism (SNP) has been observed in the SULT1A1 gene that results in an arginine-to-histidine amino acid change in codon 213. Individuals with 2 His²¹³ alleles have only 15% of the SULT1A1 activity compared with the carriers of the Arg²¹³ allele (14). Dietary polyphenols, such as resveratrol, may compete with PhIP as substrates and/or inhibitors for human SULT1A1. Because SULT1A1 is involved in the chemical activation of PhIP by forming highly reactive ester compounds that form DNA adducts, resveratrol could potentially reduce cancer risk through this competitive inhibition. The SULT1A1 codon 213 variant, which has been linked with increased risk to some cancers (15), has lower activity toward both resveratrol (16) and PhIP (13). PhIP and its carcinogenic metabolite N-hydroxy-PhIP (N-OH-PhIP) are extensively conjugated by UDP-glucuronosyltransferase (UGT). UGT1A1 is the predominant UGT involved in PhIP metabolism, and genetic variation in UGT1A1 may affect the rate of PhIP metabolism (17, 18).

On the basis of the potential for PhIP metabolism to be affected by the chemical compounds in common dietary beverages, we hypothesized that beverage consumption patterns may influence prostatic levels of PhIP–DNA adducts. Therefore, we evaluated potential associations between beverage intake and PhIP–DNA adduct levels (as quantified by immunohistochemistry) in 391 prostate cancer patients who underwent radical prostatectomy, taking into account variation in the SULT1A1 and UGT1A10 genes and the level of African ancestry.

The study population consisted of men who were part of the Henry Ford Health System. Details about the ascertainment and recruitment of study cases can be found in a previous publication (19). This study includes 391 of the 419 prostate cancer cases that underwent radical prostatectomy and had tissue specimens with sufficient areas of tumor and nontumor cells for immunohistochemical DNA adduct studies. The mean age at diagnosis was 60.8 ± 6.6 years with an average of 2.9 ± 2.4 months between diagnosis and surgery. The 391 study participants that comprised the analytic sample self-identified as 56.5% white and 43.5% African American. Cases had a mean age of 60.8 ± 6.7 years at time of diagnosis with 42.7% of patients diagnosed under the age of 60 years. The majority of patients (80.1%) had PSA levels greater than 4 ng/mL at diagnosis with a median PSA level at 5.2 ng/mL. A Gleason grade of 7 was the predominant tumor grade (45.9%) of cases with 19% of cases having a Gleason grade of 8 or higher. Most cases (80%) had stage 2 tumors with a median tumor volume of 15% of the prostate gland.

Beverage consumption was measured by self-report, using the SELECT food frequency questionnaire (FFQ) developed from FFQs used in the Women's Health Initiative (20) and the Prostate Cancer Prevention Trial (21). The FFQ asked about usual beverage consumption over the year preceding prostate cancer diagnosis and included portion and frequency information for 17 beverages or beverage additives, such as milk in coffee or tea. The distribution of amount consumed for any of the specific beverages was not uniform; therefore, we limited our analysis to a yes/no variable as to whether a beverage was consumed for the 15 beverage items with consumption frequencies. Because the SELECT questionnaire does not include food preparation questions, supplemental questions adapted from a validated questionnaire were added to assess grilled meat intake (22). To account for dietary PhIP exposure relevant to PhIP–DNA adduct levels in prostate, as previously reported by our group (23), in multivariate analyses we adjusted for combined red and white grilled meat consumption, using a dichotomous dummy variable defined as 0 to 2 servings and more than 2 servings per month.

Hematoxylin and eosin–stained slides of study cases were reviewed by the study uropathologist (A.T. Savera) to confirm the diagnosis and identify a paraffin block with sufficient tumor and nontumor prostatic tissue for staining. Using a microtome, 5 consecutive sections (5-μm thick) were cut from the tissue block of each patient sample. One slide was hematoxylin and eosin stained and examined by the study uropathologist who circled separate areas of tumor and nontumor cell populations to be used for adduct scoring. The paraffin-embedded sections were baked at 50°C for 1 hour, deparaffinized in xylene, and rehydrated in serial alcohol. After treatment with RNase and proteinase K, the sections were blocked by using 3% BSA and normal goat serum. The primary anti–PhIP–DNA adduct polyclonal antibody (24, 25) was incubated with the sections at 4°C overnight in a humid chamber at a dilution of 1:500. Also, the biotinylated secondary antibody was incubated with the sections at room temperature for 30 minutes, at a dilution of 1:200. Endogenous peroxidase activity was blocked by using 0.3% hydrogen peroxide in methanol for 20 minutes. The antibody complex was detected by an avidin–biotin–peroxidase complex solution and visualized by 3,3′-diaminobenzidine (Zymed Laboratories, Inc.). A negative control was included in each experiment by omitting the primary antibody. The staining specificity was confirmed by using the primary antibody that had been preabsorbed with 2 or 20 μg/mL DNA extract from MCF-7 cells treated with 150 μmol/L N-OH-PhIP. A cytospin sample of MCF-7 cells without PhIP treatment was included in each batch of staining. Staining was measured by absorbance image analysis, using a Cell Analysis System 200 microscope. Absorbance of light at a wavelength of 500 nmol/L was measured because methyl green does not absorb light at this wavelength, whereas diaminobenzidine does. For each prostate specimen, 2 technicians independently scored 50 epithelial cells (5 fields with 10 cells per field scored) in the 2 areas (tumor and nontumor) circumscribed by the study pathologist. The final score was the mean of the 2 technicians' scores. Scored cells were selected to be representative, in terms of intensity, of the cells in the field. Staining intensity was represented by the absorbance unit of optical density.

Genotype data for this study were generated by using different methods. Most genotypes came from 2 separate Oligo Pool All (OPA) run on the Illumina GoldenGate genotyping platform. The first OPA was a panel of 1,509 SNPs informative for West African versus European ancestry that was a precursor to the current Illumina SNP panel for determining ancestry (http://www.illumina.com/products/african_american_admixture_panel.ilmn). The second OPA was a custom panel of 1,473 tag SNPs for 172 candidate genes drawn from the phase II version of the International HapMap project (http://www.hapmap.org). To optimally select a minimal number of tag SNPs that capture variation in both of the European and African samples, we used the multiple population tagging method TAGster (26). For the purposes of this study, we used genotype data on three tag SNPs (genotyping of 4 tag SNPs attempted) for the SULT1A1 gene and 59 (successfully genotyped 60 of 66 attempted SNPs with one SNP removed because it had an exact Hardy-Weinberg equilibrium P < 0.01) tag SNPs for the UGT1A10 gene that tagged approximately 80%/98% of the common (10% frequency or greater) variation present in these two genes in the HapMap YRI/CEU samples (see the Supplementary Methods for more details). We also genotyped the known functional exon 7 SNP of the SULT1A1 gene (rs9282861) that results in an amino acid change (Arg to His) at codon 213 of the translated protein. The PCR reactions were based on a previous report (27) and performed in a total volume of 25 μL of a solution containing 5× Colorless GoTaq Flexi Buffer, 8% dimethyl sulfoxide, 0.2 mmol/L deoxynucleotide triphosphate, 2 mmol/L MgCl2, 0.8 μmol/L of each primer (upstream primer 5′-GGTTGAGGAGTTGGCTCTGC-3′ and downstream primer 5′-ATGAACTCCTGGGGGACGGT-3′), and 1.25 U of GoTaq DNA Polymerase. The reaction started with 95°C for 2 minutes, followed by 40 cycles consisting of denaturation (94°C for 1 minute), annealing (60°C for 1 minute), and extension (72°C for 2 minutes). The 289-bp PCR product was digested with the restriction enzyme HhaI at 37°C for 4 hours.

χ² tests were used to evaluate differences in beverage intake by race with a statistical correction made for multiple comparisons (28). Potential batch effects in the PhIP–DNA adduct assay were taken into account by computing a batch correction factor that was the difference between the adduct level of the positive control slide in a single batch and the mean adduct level of the positive control slides across all batches. The batch-adjusted adduct level was the crude adduct level minus the batch correction factor. To account for correlation of observations within the same experimental batch, we used a mixed linear model implemented as PROC MIXED in SAS (29). In this modeling framework, the correlations of observations within each experimental batch were treated as random effects whereas other model covariates were treated as fixed effects.

Principal components analysis (PCA) was used to derive linear transformations of the original tag SNPs data, in which eigenvectors are chosen to maximize the variance of each PC relative to the overall variation in the region (30). The number of principal components needed for each gene was determined by using an 80% explained variance criteria, reducing the number of parameters to be tested. Once the necessary PCs were determined, generalized linear models were used to assess the significance of the gene. Please see the Supplementary Methods section for a more detailed description of how PCs were determined and loaded on individual SNPs. For a subset of the African American cases that were genotyped with ancestry informative markers, estimates of African ancestry were generated with the ADMIXMAP statistical program (http://www.homepages.ed.ac.uk/pmckeigu/admixmap/), which uses a hybrid of Bayesian and traditional modeling to compare observed versus expected ancestry across the genome (31).

Among the 391 cases with PhIP–DNA adduct assessment, on average 7.5 different beverages were consumed (of the 15 types of beverages queried) with 7 types of beverages consumed the mode and median response. Overall, cases in the study averaged 7.4 beverage servings per day. Table 1 shows how reporting of consumption of specific beverages varied by race. African Americans and whites did not differ in the percentage of men consuming milk, tea, tomato juice, citrus juice, or liquor, but a significantly lower percentage of African American men reported consuming coffee (71.2% vs. 88.2%, P < 0.001), beer (34.1% vs. 56.6%, P < 0.001), diet pop (27.1% vs. 47.1%, P < 0.001), and wine (21.8% vs. 44.3%, P < 0.001) compared with white men. A higher percentage of African American cases reported consuming other juices (77.6% vs. 53.8%, P < 0.001), fruit juices (48.1% vs. 9.5%, P < 0.001), regular pop (73.5% vs. 52.9%, P < 0.001), and meal replacement (9.4% vs. 3.6%, P = 0.02).

Mean PhIP–DNA adduct levels by type of beverage consumed are reported in Table 2. Mean values are reported separately for tumor and adjacent nontumor tissue; these were adjusted for race, age, grilled meat consumption, Gleason score, and tumor volume, which are the 5 factors we found previously to be associated with PhIP–DNA adduct levels in prostate (8, 23). In prostate tumor cells, significantly lower mean PhIP–DNA adduct levels were found among cases that reported consuming red wine (P = 0.006) and any wine (P = 0.04). Cases who consumed white wine or beer also had lower PhIP–DNA adduct levels in tumor, but the P value for these differences were not significant (P = 0.09 for both). In nontumor cells, the same trends were observed between consumers and nonconsumers. In prostate nontumor cells, red wine consumption showed the strongest inverse relationship with PhIP–DNA adduct levels. After correcting for multiple comparisons within the 2 cell types, the red wine inverse association in prostate nontumor cells was the only association that remained statistically significant at the 0.05 alpha level (P_corrected = 0.032). We found no differences in adjusted PhIP–DNA adduct levels between whites and African Americans (0.154 ± 0.015 vs. 0.156 ± 0.015; P = 0.6), and race-stratified comparisons showed similar beverage consumption associations compared with the analysis of the full sample.

We next conducted analyses stratified by red wine and grilled red meat consumption and race to discern their joint effects on PhIP–DNA adduct levels in prostate (Fig. 1). In nontumor cells, the African American and white cases with high grilled red meat consumption that did not drink red wine had 23% (P = 0.009) and 16% (P = 0.01) higher PhIP–DNA adduct levels compared with their respective race matched counterparts with low grilled meat consumption who drank red wine. In African American cases, the inverse association between red wine exposure and PhIP–DNA adduct levels in nontumor cells was only observed in cases with low grilled red meat exposure, whereas in whites red wine consumption was inversely associated with PhIP–DNA adducts in both the high and low grilled red meat exposure groups. In tumor cells, the African American and white cases with high grilled red meat consumption that did not drink red wine had 18% (P = 0.06) and 22% (P = 0.005) higher PhIP–DNA adduct levels than low grilled meat consumers who drank red wine.

To determine how genetic factors related to the metabolism of PhIP might affect the relationship between PhIP–DNA adducts and red wine consumption, we tested several race- and tissue type–specific multivariable models that included covariates for genetic factors potentially associated with PhIP–DNA adducts (Table 3). The UGT1A10 and SULT1A1 PCA factors were derived from race-specific principal components models that tested for the number of principal variance components that would account for 80% or more of the variation within the UGT1A10 and SULT1A1 tag SNPs genotyped. For whites, 80.4% of UGT1A10 common variation was described by 6 principal components (denoted PCA1…PCA6). For African Americans, 81.2% of UGT1A10 common variation was described by 12 principal components (denoted PCA1…PCA12). For SULT1A1 variation, 93.0% of the common variation was described by the first 2 of 3 principal components in whites, whereas in African Americans the first 3 of 4 principal components described 90.1% of SULT1A1 common variation. These genetic factors included 2 race-specific UGT1A10 PCA factors that were most strongly associated with PhIP–DNA adduct levels and 1 SULT1A1 PCA factor that showed an association with PhIP–DNA adduct levels in African American cases.

In whites, a significant negative association with red wine consumption was observed in both nontumor and tumor cells that increased by approximately 30% after adjusting for the 2 UGT1A10 principal components most strongly associated with PhIP–DNA adduct levels. In a multivariable model of PhIP–DNA adduct levels in tumor cells, the association of red wine consumption was significant at the P = 0.01 level. In African American cases, before adjusting for genetic factors the beta coefficients for red wine consumption in nontumor cells were of similar magnitude and direction as was observed in whites, although of lower significance likely because of the smaller sample size. However, unlike in white cases, in African American cases adjustment for genetic factors left the effect estimates for red wine consumption unchanged in both nontumor and tumor cells. The additional covariate of African ancestry showed a statistically significant (P = 0.004) association with PhIP–DNA adduct in nontumor cells. Specific genetic factors were also more prominently associated with PhIP–DNA adduct levels in African American cases, most notably the third principal component for SULT1A1 common variation and the fourth principal component for UGT1A10 common variation.

PhIP–DNA adducts are a measurable biological effective dose of dietary heterocyclic amine exposure (32). Having previously shown an association between consumption of grilled red meat and PhIP–DNA adduct levels in human prostate (23), in this study we tested for associations between consumption of 15 different beverages and PhIP–DNA adduct levels in prostate. Red wine consumption had the strongest association with PhIP–DNA adduct levels being inversely related to adduct levels in both prostate tumor and nontumor cells.

Recently, red wine and its most biologically active component resveratrol, a polyphenolic phytoestrogen found in high concentrations in the skins of grapes used to make red wine (33), has drawn attention as a potential chemopreventive agent for prostate cancer (34). Epidemiologic support for the chemopreventive effects of red wine on prostate cancer comes mainly from a case–control study that showed an inverse dose response with average glasses of red wine consumed per week and risk of prostate cancer (35); however, cohort studies have failed to replicate this finding (36–38). Resveratrol has generated considerable research interest because of its potential wide range of biological effects that include antioxidant, chemopreventive, cardioprotective, neuroprotective, anti-inflammatory, and antiviral activities (39, 40). Much of the research into the potential biological mechanisms that underlie resveratrol's chemopreventive effects has focused on its antioxidant capacity, but one proposed mechanism that may have relevance in relation in prostate cancer is resveratrol's potency in decreasing the mutagenic potential of food-derived heterocyclic amines such as PhIP and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline at micromolar concentrations (41). One possible explanation for the inhibitory effects of resveratrol on PhIP mutagenicity is that the activation of PhIP to its reactive DNA binding species and conjugation of resveratrol, resulting in its reduced bioavailability, both occur through the same sulfation enzymatic reaction (42, 43). In primary cultures of human mammary epithelial cells, administration of resveratrol inhibited sulfotransferase activity twice as effectively as it inhibited acetyltransferase activity and reduced PhIP–DNA adduct levels to approximately 50% of that observed in untreated control cells (11). Furthermore, the most common functional genetic variant of the SULT1A1 gene, the codon 213 His/Arg substitution, affects the enzymatic activity of this key sulfotransferase protein toward both PhIP (13) and resveratrol (16).

In humans, PhIP detoxification occurs primarily in the liver by glucuronidation of PhIP and N-OH-PhIP by UGT with increased levels in urine of the main metabolite, N-OH-PhIP-N²-glucuronide, corresponding to lower levels of DNA adducts found in the colon of individuals exposed to PhIP (44). The UGT superfamily of enzymes are coded by the UGT1 locus on chromosome 2q37 that consists of a set of common exons in the 3′ region designated 2, 3, 4, and 5 and an exon 1 unique to each gene in the UGT family (45). UGT enzymes catalyze the glucuronidation of a variety of endogenous compounds and are integrally involved in the detoxification of many carcinogens, the clearance of drugs, and the metabolism of a variety of endogenous compounds (46). Among the UGT enzymes, UGT1A10 has exhibited significantly higher glucuronidation rates against PhIP and N-OH-PhIP than any other UGT family member (18).

Sulfation plays a key role in the activation of N-hydroxy derivatives of carcinogenic heterocylica amines, such as PhIP (47). A role for SULT1A1 in PhIP-induced prostate carcinogenesis is biologically plausible because the SULT1A1 transcript has been detected in human prostate (48). In studies using functional recombinant proteins from cDNA inserted in Salmonella Typhimurium strain, N-OH-PhIP was activated specifically by human SULT1A1, but not by NAT1 or NAT2 (49). In humans that ingested a radioactively labeled PhIP compound, SULT1A1 activity is positively correlated with PhIP–DNA adduct formation in colon and a much stronger predictor of DNA adduct levels than NAT2 activity (44), which would be the alternative activation pathway of N²-OH-PhIP. Similar results were found in PhIP exposure experiments conducted with prostate cytosols prepared from benign prostate tissue collected from men undergoing transuretheral resections of the prostate in that SULT1A1 protein levels, but not NAT2 protein levels, were correlated with DNA adduct levels (50). SULT1A1-catalyzed sulfation of PhIP results in an unstable sulfoxy metabolite of N-OH-PhIP that readily binds to DNA, or a 4′-SO4-PhIP metabolite that is excreted in the urine (51). Although SULT1A1 can then both potentially activate and detoxify PhIP, making it difficult to discern the overall role of SULT1A1 in PhIP-induced carcinogenesis, clearly SULT1A1 activity is influenced by resveratrol and plays a role in DNA adduct formation.

We found that the level of African ancestry was strongly associated with PhIP–DNA adduct levels in nontumor prostate cells. Nowell and colleagues found a more pronounced association with increasing SULTA1 activity and prostate cancer risk in African Americans compared with whites (52). Common genetic variation is known to be associated with gene expression levels (53) and recent work suggests that gene expression differences between African American and European American tumors may lead to group-specific alterations in cancer-related pathways that control host response, inflammation, and tumor angiogenesis (54). In light of these findings, we hypothesize that our “African ancestry” variable may be a surrogate for differential gene expression that could influence how PhIP is detoxified by UGTs as well as the effects that resveratrol and SULT1A1 have on N²-OH-PhIP activation (Fig. 2).

In summary, we have found red wine consumption to be inversely associated with PhIP–DNA adduct levels in prostate tumor and nontumor cells of men with prostate cancer. Because PhIP is known to be a potential prostate carcinogen, red wine may hold renewed promise as a chemopreventive agent for prostate cancer assuming that (a) PhIP exposure that results in DNA damage through adduct formation significantly increases prostate risk and (b) the active agent in red wine, resveratrol, through competitive inhibition with SULT1A1 (or some other yet to be defined mechanism) can significantly reduce the carcinogenic effects of PhIP exposure in human prostate. The first point has yet to be proven, but is an area of active investigation for our group. Previous work in controlled experimental systems (11) and our results suggest the second point is plausible, but given the cross-sectional nature of our results, further studies where the dose and timing of resveratrol consumption with respect to PhIP exposures can be controlled are needed to validate how resveratrol acts on PhIP metabolism in humans. In conclusion, our results suggest that the effects of PhIP exposure can be mitigated by red wine consumption, but before any chemoprevention efforts are directed toward this pathway, the influence of inherited genetic factors that vary by race on PhIP metabolism needs to be better understood.

No potential conflicts of interest were disclosed.

This work was supported by the National Institute of Environmental Health Sciences R01 ES11126 grant awarded to B.A. Rybicki and Department of Defense grant PC050775 awarded to C.H. Bock.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.