Interactions Between CYP2C9 and UGT1A6 Polymorphisms and Nonsteroidal Anti-Inflammatory Drugs in Colorectal Cancer Prevention

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2006;4: Interactions Between CYP2C9 and UGT1A6 Polymorphisms and Nonsteroidal Anti-Inflammatory Drugs in Colorectal Cancer Prevention WADE S. SAMOWITZ,* ROGER K. WOLFF, KAREN CURTIN, CAROL SWEENEY, KHE NI MA, KRISTEN ANDERSEN, THEODORE R. LEVIN, and MARTHA L. SLATTERY *Department of Pathology, and the Health Research Center, University of Utah Health Sciences Center, Salt Lake City, Utah; University of Minnesota, Minneapolis, Minnesota; and the Kaiser Permanente Medical Research Program, Oakland, California Background & Aims: Variant genotypes of uridine diphosphate glucuronsyltransferase isoenzyme 1A6 (UGT1A6) associated with decreased metabolic activity have been associated with an enhanced protective effect of aspirin on the development of colorectal adenomas. However, interactions between UGT1A6 variants or variants of another enzyme that metabolizes nonsteroidal anti-inflammatory drugs (NSAIDs), cytochrome P4502C9 (CYP2C9), and NSAIDs in the prevention of colorectal cancer have not been studied extensively. Methods: UGT1A6 and CYP2C9 genotypes were determined in 2295 individuals with colorectal cancer and 2903 controls. Interactions between these genotypes, aspirin or ibuprofen use, and colorectal cancer risk were determined. Results: Variant CYP2C9 genotypes enhanced the protective effect of ibuprofen on the prevention of colorectal cancer, and a dose-response relationship with respect to increasing numbers of variant alleles was seen (P interaction.02). CYP2C9 variants were more effective in individuals with wild-type rather than variant UGT1A6 (P interaction <.007). Variant CYP2C9 genotypes showed no interaction with aspirin usage, and variant UGT1A6 genotypes showed no interaction with either NSAID with respect to colorectal cancer protection. Conclusions: In this study, the major effect seen was an enhancement by slower-metabolizing CYP2C9 variants of the chemopreventive activity of ibuprofen against colorectal cancer. One of the most reproducible findings in studies of colorectal cancer prevention is the protective effect of aspirin and other nonsteroidal anti-inflammatory drugs (NSAIDs). 1 NSAIDs are metabolized primarily by the enzymes uridine 5=-diphosphate glucuronosyl transferase 1A6 (UGT1A6) and cytochrome P4502C9 (CYP2C9), and there is evidence that UGT1A6 is involved predominantly in the breakdown of aspirin whereas CYP2C9 is involved predominantly in the metabolism of nonaspirin NSAIDs. 2,3 Both of these enzymes have polymorphisms that substantially reduce their catabolic efficacy. With respect to UGT1A6, a T181A/R184S double polymorphism has a reported allele frequency of 30%, whereas the single R184S has a reported frequency of 2%; these amino-acid changes result in enzymes with 30% 50% of the activity of the wild-type form. 4,5 The CYP2C9 polymorphisms R144C and I159L have reported allele frequencies of approximately 10% and 6%, respectively, and are associated with 5% 30% of the function of the wild-type enzyme. 6 9 A reasonable hypothesis is that the decreased NSAID metabolism associated with these polymorphisms might intensify the protective effect of NSAIDs on colorectal cancer. Several studies have analyzed the influence of these polymorphisms on the development of the precursor of most colorectal cancers, the adenomatous polyp. Both Bigler et al 8 and Chan et al 4 found a protective effect of UGT1A6 variants on adenomatous polyp formation among those who used aspirin; indeed, the Chan et al 4 study, which was restricted to women, implied that only those with the UGT1A6 variants benefited from aspirin therapy. With respect to CYP2C9, another study by Chan et al 7 found no interaction between CYP2C9 polymorphisms and aspirin use with respect to the prevention of distal adenomatous polyps, whereas the study by Bigler et al 8 reported the somewhat counterintuitive result of adenoma prevention only with the wild-type CYP2C9 allele. Abbreviations used in this paper: CYP2C9, cytochrome p450 2C9; KPMCP, Kaiser Permanente Medical Care Program of Northern California; NSAIDs, nonsteroidal anti-inflammatory drugs; PCR, polymerase chain reaction; UGT1A6, uridine diphosphate glucuronsyltransferase isoenzyme 1A6; var, variant; wt, wild type by the American Gastroenterological Association Institute /06/$32.00 doi: /j.cgh

2 July 2006 CYP2C9, UGT1A6, AND COLORECTAL CANCER 895 Although studies on adenomas certainly are interesting, the effects of these polymorphisms on the development of colorectal cancer itself is, perhaps, more clinically relevant. To date, only 1 study 10 has examined interactions between NSAIDs and polymorphisms in these enzymes in the prevention of colorectal cancer, and that study did not find any such interaction. We previously have used a large population-based study of individuals with colorectal cancer to show the protective effects of aspirin and ibuprofen on the development of colorectal cancer, although the effect of aspirin on rectal cancer was seen only in women. 11 We now use this sample in an attempt to resolve the apparent discrepancy between studies of colorectal adenomas and carcinomas with respect to whether interactions exist between aspirin and/or ibuprofen and polymorphisms in UGT1A6 and CYP2C9 in the prevention of colorectal neoplasia. Methods Study Populations Participants were from the Kaiser Permanente Medical Care Program of Northern California (KPMCP), the state of Utah, and the Twin City Metropolitan area of Minnesota (colon cancer study only). All eligible patients within these defined areas were identified and recruited for a study of colon cancer and a study of rectal cancer. The first study included patients and controls from a populationbased case-control study of first primary colon cancer (International Classification of Diseases of Oncology 2nd edition codes 18.0 and ) diagnosed between October 1, 1991, and September 30, 1994, conducted in all 3 target areas. Patients from the second study were diagnosed with a first primary tumor in the rectosigmoid junction or rectum and were identified between May 1997 and May 2001, and were restricted to patients and controls from Utah and KPMCP. Patient eligibility was determined by the Surveillance Epidemiology and End Results Cancer Registries in northern California and in Utah and the Minnesota Surveillance System (colon cancer patients only). In both studies, patients were identified using rapid-reporting systems. For both studies, eligibility included age between 30 and 79 years of age at time of diagnosis, English speaking, mentally competent to complete the interview, no previous history of colorectal cancer, and no known (as indicated on the pathology report) familial adenomatous polyposis, ulcerative colitis, or Crohn s disease. 12 For the colon cancer study, the median time from diagnosis to interview was 131 days overall (126 days at KPMCP, 154 days in Minnesota, and 109 days in Utah). The median time from diagnosis to interview was longer for the rectal cancer study, primarily because of the different levels of permission needed before contacting patients; at KPMCP the median time from diagnosis to interview was 154 days and for Utah was 183 days. Of colon cancer patients contacted, 83% participated at KPMCP, 76% in Utah, and 67% in Minnesota. For the rectal cancer study, the cooperation rates were 75.4% of patients from KPMCP and 69.7% of patients from Utah. Controls were frequency matched to patients by sex and by 5-year age groups, but not by race. At KPMCP, controls were selected randomly from membership lists. In Utah, controls age 65 years and older were selected randomly from lists provided by the Centers for Medicare and Medicaid Services (formerly Health Care Finance Administration [HCFA]) and controls younger than age 65 were selected randomly from driver s license lists. In Minnesota, controls were selected randomly from driver s license lists. Of controls contacted for the colon cancer study, 73% participated at KPMCP, 53% participated from Minnesota, and 69% participated from Utah. For the rectal cancer study, cooperation rates were 70% for KPMCP and 67% for Utah. Data were collected by trained interviewers; quality control procedures have been described previously. 13 The referent period was the calendar year 2 years before the date of diagnosis (patient) or selection (controls). Regular use of aspirin and ibuprofen was defined as use at least 3 times a week for 1 month or more within the past 2 years. Genotyping DNA was extracted from peripheral blood leukocytes. For quality control, known controls representing all polymorphic variants and blanks (negative controls) were included in each 96-well tray. All genotypes were scored by 2 individuals. Genotyping Cytochrome P4502C9 R144C and I359L. Two independent assays were performed to assess the 2 variant alleles of CYP2C9. The R144C polymorphic variant that defines CYP2C9*2 was genotyped using a polymerase chain reaction (PCR)-restriction fragment length polymorphism approach. 14 PCR amplification was performed using 20 ng of genomic DNA, primers CYP2C9a2-F (5=-GTA TTT TGG CCT GAA ACC CAT A-3=) and CYP2C9a2-R (5=-GGC CTT GGT TTT TCT CAA CTC-3=), 1.5 mmol/l MgCl 2, deoxynucleoside triphosphates, and.25 units of Amplitaq polymerase (Applied Biosystems, Foster City, CA). After an initial denaturation at 95 C for 2 minutes, the samples were subjected to 35 cycles consisting of 95 C for 20 seconds, 60 C for 30 seconds, and 72 C for 40 seconds. PCR products then were subjected to restriction digest with AvaII according to the manufacturer s (NEB, Ipswich, MA) suggested conditions. The AvaII digested samples then were size fractionated on agarose gels containing ethidium-bromide and visualized with ultraviolet light. Products that remained uncut (454 bp) correspond to allele *2 (C144 allele), and samples containing bands migrating at 397 bp and 57 bp (cut) are either allele *1 or *3 (R144 containing alleles). The CYP2C9 I359L polymorphism was genotyped using a TaqMan-based assay as described by Chan et al. 7 We used

3 896 SAMOWITZ ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 7 primers CYP2C9-359-F (5=-GCC ACA TGC CCT ACA CAG ATG-3=) and CYP2C9-359-R (5=-GAA TTT AAT GTC ACA GGT CAC TGC AT-3=) with probes CYP2C9 I359 (5=-VIC- AGG TCA ATG TAT CTC T-Tamra-3=) and CYP2C9 L359 (5=-FAM-AAG GTC AAG GTA TCT C-Tamra-3=). Each 17 ul TaqMan reaction contained 20 ng genomic DNA, 900 nmol/l of each primer, 130 nmol/l of each TaqMan probe, and 8.5 ul TaqMan Universal PCR Master Mix (contains Amp- Erase UNG, Applied Biosystems, Foster City, CA, and AmpliTaq Gold enzymes, deoxynucleoside triphosphates, and reaction buffer). PCR was performed under the following conditions: 50 C for 2 minutes (UNG activation), 95 C for 10 minutes, followed by 40 cycles of 92 C for 15 seconds, and 60 C for 1 minute using the Bio-Rad IQ detection system (Bio-Rad, Hercules, CA). The fluorescence of each sample was collected and analyzed by version 3.0 of the icycler IQ Real- Time detection software (Bio-Rad). The I359 variant is present in both alleles *1 and *2, and the L359 variant is present only in allele *3. In summary, the CYP2C9 genotypes were scored as follows: *1: R144/I359; *2: C144/I359; *3: R144/L359. Genotyping Uridine 5=-Diphosphate Glucuronyl Transferase 1A6 T181A and R184S. UGT1A6 variant alleles also were evaluated by 2 independent assays. The T181A polymorphic variant was genotyped using a PCR restriction fragment length polymorphism method. 15 PCR amplification was performed using 20 ng genomic DNA, primers UGT1A6-F (5=-GGA AAT ACC TAG GAG CCC TGT GA-3=) and UGT1A6-R (5=-AGG AGC CAA ATG AGT GAG GGA G-3=), 1.5 mmol/l MgCl 2, deoxynucleoside triphosphates, and.25 units of Amplitaq polymerase. After an initial denaturation at 95 C for 2 minutes, the samples were subjected to 35 cycles consisting of 95 C for 30 seconds, 64 C for 45 seconds, and 72 C for 45 seconds. PCR products then were subjected to restriction digest with NsiI according to the manufacturer s (NEB) suggested conditions. The NsiI digested samples then were size fractionated on agarose gels containing ethidium bromide and visualized with ultraviolet light. The T181 variant gave a product of 992 bp (uncut by NsiI) and the A181 variant gave products of 616 bp and 376 bp (NsiI cut). The UGT1A6 R184S variant was genotyped using a Taq- Man-based assay. The TaqMan assay was generated using Assay by Design (Applied Biosystems). The assay s ability to discern alleles for UGT1A6-R184S was confirmed by comparison of results based on PCR restriction fragment length polymorphism analysis using enzyme Fnu4HI 15 (data not shown). We used primers UGT1A6-184F (5=-CCG TGT TCC CTG GAG CAT-3=) and UGT1A6-184R (5=-CAC CTG GGA ATG TAG GAC ACA-3=) with probes UGT1A6-R184 (5=- VIC-TTC AGC AGA AGC CCA G MGBNFQ-3=) and UGT1A6-S184 (5=-FAM-CAG CAG CAG CCC AG MGB- NFQ-3=). Each 20 ng of genomic DNA was amplified in 17 L reaction containing 8.5 ul of 2 TaqMan Universal PCR Master Mix, and.18 ul of 80 Assay by design mix. PCR was performed using the Bio-Rad icycler with the following conditions: 50 C for 2 minutes (UNG activation), 95 C for 10 minutes, followed by 40 cycles of 95 C for 15 seconds, and 60 C for 1 minute. The fluorescence of each sample was collected and analyzed using version 3.0 of the icycler IQ Real-Time detection software. Three alleles are present in the population for UGT1A6. They are T181 and R184 (wild type, *1) and variant alleles A181 and S184 (*2), and T181 and S184 (*3). Diplotypes and Variant Alleles The following diplotypes and the respective number of variant alleles were observed for CYP2C9 and UGT1A6 (see previously for definition of *1, *2, and *3 for each gene: *1/*1: no variant alleles; *1/*2: 1 variant allele; *1/*3: 1 variant allele; *2/*2: 2 variant alleles; *2/*3: 2 variant alleles; and *3/*3: 2 variant alleles. Statistics Unconditional logistic regression models were used to estimate the relative risk of colon and rectal cancer from aspirin, ibuprofen, UGT1A6 variants, and CYP2C9 variants. We evaluated associations between aspirin and ibuprofen for regular use within the 2 years before diagnosis or selection. In these models the following variables were included as adjustment variables because they have been shown previously to be associated with colorectal cancer risk in this population: age at selection, sex, body mass index, physical activity, energy intake, usual number of cigarettes smoked per day, and dietary calcium and fiber intake. Further adjustment for family history and hormone use had no effect on the results. The interaction between genotype and ibuprofen and aspirin was assessed using the following methods: multiplicative interaction was determined by the cross-product of the genotype and aspirin or ibuprofen; the SAS statistical package (version 8.2; Cary, NC) was used to conduct the analyses; SAS Genetics was used to test for Hardy Weinberg equilibrium and determine allele frequencies. We addressed potential confounding effects of other risk factors on the association between aspirin or ibuprofen and colon or rectal cancer by including variables in the regression models for the following exposures: sex, age, body mass index, long-term vigorous physical activity, energy intake, average number of cigarettes per day, and dietary calcium and fiber intake. All of these covariates, except sex, were entered into the model as continuous variables. The same set of adjustment variables was used for all analyses. We assessed interaction between genotype and the use of aspirin or ibuprofen on a multiplicative scale by testing for significance of a term representing the cross-product of aspirin or ibuprofen (dichotomous, yes or no) and genotype (ordered categories of number of variant alleles, treated as a continuous variable).

4 July 2006 CYP2C9, UGT1A6, AND COLORECTAL CANCER 897 Table 1. Description of Study Population Colon cancer Rectal cancer Patients % Controls % Patients % Controls % Age (y) Ethnicity White Hispanic Black Asian American Indian, Alaskan Native Sex Male Female Center Utah KPMCP Minnesota Family history of colorectal cancer No Yes Results A general description of the study population is shown in Table 1. The allele frequencies for the UGT1A6 T181A/R184S double polymorphism in colorectal study controls were.31 in whites,.26 in Hispanics, and.29 in blacks, and the frequencies for the R184S single polymorphism were.04,.10, and.06, respectively, for whites, Hispanics, and blacks. The allele frequencies for the CYP2C9 R144C and I359L were.13 and.06 in whites,.15 and.02 in Hispanics, and.04 and.02 in blacks, respectively. Individual UGT1A6 polymorphisms and CYP2C9 polymorphisms were in Hardy Weinberg equilibrium in controls (data not shown). The patient and control populations in this study were predominantly white (88% and 91%, respectively). Hispanics comprised 5% of patients and 4% of controls, and blacks comprised 4% of patients and 3% of controls. Adjustment for ethnicity did not affect the results described later (data not shown). UGT1A6 and CYP2C9 variant genotypes by themselves were not associated with an increased or decreased risk of colon or rectal cancer (Table 2). However, the use of either aspirin or ibuprofen was associated with a decreased risk of both colon and rectal cancer (although the protective effect of aspirin against rectal cancer was seen only in women, as reported previously for these study populations). 11 Colorectal Cancer, Genetic Polymorphisms, and Nonsteroidal Anti-Inflammatory Drug Use The presence of any CYP2C9 variant genotype enhanced the protective effect of ibuprofen against colorectal cancer, and a dose-response relationship was evident with increasing numbers of variant alleles (P trend.02, Table 3). This effect was not seen with aspirin and CYP2C9 genotypes, nor was there any interaction between UGT1A6 genotypes and the effects of either aspirin or ibuprofen on the risk of colorectal cancer. No significant differences were seen with respect to cancer site or sex (data not shown). Combined Uridine 5=-Diphosphate Glucuronyl Transferase 1A6/Cytochrome P4502C9 Genotypes There was a significant multiplicative interaction between ibuprofen use and the combined UGT1A6/ CYP2C9 genotypes for colorectal cancer because variant CYP2C9 genotypes were more effective in individuals who were UGT1A6 wild type rather than variant (P interaction.007, Table 4). The various genotype combinations did not interact with the effect of aspirin on the development of colorectal cancer.

5 898 SAMOWITZ ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 7 Table 2. Associations of UGT1A6, CYP2C9, Aspirin, and Ibuprofen With Colon and Rectal Cancer Colon cancer Rectal cancer Cases Controls OR Cases Controls OR N(%) N (%) (95% CI) N (%) N (%) (95% CI) UGT1A6 1/ (40.4) 813 (41.9) 1.00 referent 296 (44.1) 364 (42.3) 1.00 referent 1/ 2 or 1/ (45.4) 861 (44.4) 1.07 ( ) 298 (44.4) 393 (45.7).94 ( ) 2/ 2, 2/ 3, or (14.2) 265 (13.7) 1.11 ( ) 77 (11.5) 103 (12.0).92 ( ) Any 2 or (59.6) 1126 (58.1) 1.08 ( ) 375 (55.9) 496 (57.7).94 ( ) CYP2C9 1/ (67.8) 1302 (67.7) 1.00 referent 518 (67.5) 644 (65.7) 1.00 referent 1/ 2 or 1/ (29.5) 570 (29.6) 1.04 ( ) 225 (29.3) 303 (30.9).93 ( ) 2/ 2, 2/ 3, or (2.7) 51 (2.7) 1.08 ( ) 25 (3.3) 33 (3.4).93 ( ) Any 2 or (32.2) 621 (32.3) 1.04 ( ) 250 (32.6) 336 (34.3).93 ( ) Aspirin No 1568 (79.2) 1704 (71.2) 1.00 referent 722 (76.3) 859 (71.8) 1.00 referent Yes 412 (20.8) 690 (28.8).63 (.54.73) 224 (23.7) 338 (28.2).78 (.64.96) Ibuprofen No 1712 (86.6) 1952 (81.5) 1.00 referent 793 (83.5) 926 (77.4) 1.00 referent Yes 266 (13.4) 444 (18.5).65 (.54.77) 157 (16.5) 270 (22.6).65 (.52.81) NOTE. Odds ratio (OR) and 95% confidence intervals (CI) adjusted for age, sex, body mass index, physical activity, energy intake, cigarette smoking amount, and dietary calcium and fiber intake. UGT1A6 and CYP2C9 diplotypes are described in the Methods section of the text. Discussion A recent study of women found that only those with a slow metabolizing variant of UGT1A6 benefited from aspirin therapy with respect to the prevention of colorectal adenomas. 4 Because most colorectal carcinomas are thought to arise from adenomas, the implication of this study is that a simple genetic test could determine whether or not a person would benefit from aspirin therapy with respect to colorectal cancer and, possibly, with respect to other diseases affected by aspirin (eg, cardiovascular disease). That study also suggested that the CYP2C9 slower-metabolizing variants were not important in interactions with aspirin. These results were supported at least partially by a previous study of colo- Table 3. Associations Between Aspirin and Ibuprofen and the Risk of Colorectal Cancer by UGT1A6 and CYP2C9 Genotypes N patients/ controls OR (95% CI) N patients/ controls OR (95% CI) N patients/ controls OR (95% CI) OR (95% CI) UGT1A6 1/ 1 1/ 2 or 1/ 3 2/ 2, 2/ 3, or 3 3 any 2 or 3 Aspirin No 702/ referent 768/ referent 239/ referent 1.00 referent Yes 204/ (.56.84) 228/ (.55.81) 55/ (.37.79).64 (.54.76) P interaction a.39 Ibuprofen No 776/ / referent 250/ referent 1.00 referent Yes 128/ (.50.80) 157/ (.52.82) 45/67.76 ( ).68 (.56.83) P interaction a.40 CYP2C9 1/ 1 1/ 2 or 1/ 3 2/ 2, 2/ 3, or 3 3 any 2 or 3 Aspirin No 1184/ referent 526/ referent 52/ referent 1.00 referent Yes 354/ (.58.80) 135/ (.46.75) 14/26.69 ( ).59 (.47.75) P interaction a.41 Ibuprofen No 1298/ referent 566/ referent 56/ referent 1.00 referent Yes 233/ (.62.89) 101/ (.41.71) 10/25.32 (.13.78).52 (.40.68) P interaction a.02 NOTE. UGT1A6 and CYP2C9 diplotypes are described in the Methods section of the text. Odds ratio (OR) and 95% confidence interval (CI) adjusted for age, sex, body mass index, physical activity, energy intake, cigarette smoking amount, and dietary calcium and fiber intake. a From a test for interaction, on a multiplicative scale, between aspirin or ibuprofen (dichotomous, yes or no) and genotype (ordered categories of number of variant alleles, treated as a continuous variable).

6 July 2006 CYP2C9, UGT1A6, AND COLORECTAL CANCER 899 Table 4. Associations Between Aspirin and Ibuprofen and the Risk of Colorectal Cancer by Combined UGT1A6/CYP2C9 Genotypes UGT1A6 Wt UGT1A6 Var CYP2C9 Wt CYP 1/ 2 or 1/ 3 CYP 2/ 2, 2/ 3, or 3/ 3 CYP2C9 Wt CYP 1/ 2 or 1/ 3 CYP 2/ 2, 2/3, or 3/ 3 Aspirin No No. patients/controls 471/ /253 16/17 644/ /328 33/40 OR (95% CI) 1.00 referent 1.00 referent 1.00 referent 1.00 referent 1.00 referent 1.00 referent Yes No. patients/controls 146/225 46/98 6/11 190/319 78/141 7/14 OR (95% CI).71 (.56.92).55 (.36.82).69 ( ).67 (.54.83).58 (.42.81).55 ( ) P interaction a.83 Ibuprofen No No. patients/controls 516/ /264 21/16 708/ /364 32/41 OR (95% CI) 1.00 referent 1.00 referent 1.00 referent 1.00 referent 1.00 referent 1.00 referent Yes No. patients/controls 95/133 29/86 1/12 125/204 60/105 8/13 OR (95% CI).86 ( ).39 (.24.63).05 (.01.54).70 (.55.90).61 (.42.89).51 ( ) P interaction a.007 NOTE. CYP2C9 diplotypes are described in the Methods section. UGT1A6 variant refers to any genotype other than wild type. Odds ratio (OR) and 95% confidence interval (CI) adjusted for age, sex, body mass index, physical activity, energy intake, cigarette smoking amount, and dietary calcium and fiber intake. a From a test for interaction, on a multiplicative scale, between aspirin or ibuprophen (dichotomous, yes or no) and combined genotype (ordered categories treated as a continuous variable). rectal adenomas, although that study did show the paradoxic result of a beneficial effect of the wild-type CYP2C9 enzyme. 8 Factors associated with adenoma development, however, may not necessarily be associated with progression to carcinoma, and thus it is very important to determine whether the same genotype environment interaction also applied to colorectal cancer. Our results differ from these previous studies because the main effect we observed was an intensification of the protective effect of ibuprofen against colorectal cancer by CYP2C9 variant genotypes. Our results are consistent with previously reported biologic mechanisms. The strongest interaction we observed, namely that between the CYP2C9 variant alleles and ibuprofen, is consistent with previous reports that CYP2C9 is involved mostly in the metabolism of ibuprofen rather than aspirin. 2,3 Also, the enhanced effect of ibuprofen on the prevention of colorectal cancer was associated with the slower-metabolizing variants of CYP2C9, consistent with the notion that this slow metabolism would potentiate the effects of the drug. Finally, there was a dose-response relationship seen in the effects of CYP2C9 genotypes because the effectiveness of ibuprofen significantly increased with the progression from the wild type to increasing numbers of variant alleles. We also observed that slower-metabolizing variants of CYP2C9 were more effective in interacting with ibuprofen to prevent colorectal cancer in the presence of wildtype, rather than variant, UGT1A6 (Table 4). There are several plausible, albeit speculative, explanations for our observation that a UGT1A6 wild-type enzyme might confer more protection with respect to ibuprofen than the corresponding variant/polymorphic enzyme. For instance, a primary pathway of ibuprofen metabolism is the direct glucuronidation of its carboxyl group. Acyl glucuronides of ibuprofen are unstable and can react with proteins in vitro and in vivo. 16 If the reactive acyl glucuronide conjugates of ibuprofen were mechanistically relevant to cancer prevention, then a protective effect would segregate with the better enzyme (ie, the metabolically more active enzyme). In contrast, if ibuprofen itself confers protection, then the data imply that the polymorphic UGT1A6 variant is a more proficient metabolizer of ibuprofen than wildtype UGT1A6. It is notable that UGT1A6 metabolizes ibuprofen at rates of 5- to 20-fold less than UGT1A1, UGT1A3, and UGT1A9 in vitro. 17 If the polymorphism improved the ability of UGT1A6 to metabolize ibuprofen more efficiently than UGT1A6 wild type (eg, at rates comparable with UGT1A1, 1A3, and 1A9), it also could explain our findings. A slight enhancement of the protective effect of ibuprofen by wild-type UGT1A6 was seen in an evaluation of UGT1A6 by itself (Table 3), although this was not statistically significant. There are several possible reasons for discrepancies between our current study and previous studies on colorectal adenomas. Most importantly, we were studying colorectal carcinomas rather than adenomas, and factors

7 900 SAMOWITZ ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 7 associated with adenoma development may be different from those associated with the progression to cancer. The more provocative previous study 4 was limited to women, but a separate analysis of women in our population did not support their results with respect to the lack of a protective effect of aspirin against colon cancer in those who were wild type for UGT1A6 (data not shown). As in any case-control study, we could not include those who died before an interview was possible. We cannot, therefore, exclude the possibility that UGT1A6 polymorphisms and aspirin, for example, might have an effect on the incidence of rapidly fatal colon cancer. It is difficult to imagine a chemopreventive mechanism, however, in which this effect would be operative in both adenoma development (as suggested by previous studies) and in preventing the most aggressive cancers but would have no effect on other invasive cancers, many of which, although not rapidly fatal, eventually do lead to the patient s demise. A previous study that showed no interaction between these polymorphisms and NSAID use in the prevention of colorectal cancer 10 evaluated a much smaller sample than our current study and may have been relatively underpowered. There are several strengths to our study. One is that we evaluated the clinically relevant colorectal carcinoma rather than evaluating a precursor lesion as a surrogate marker for colorectal cancer. In addition, our study was population based and we therefore can extrapolate our results to the population at large. Finally, our study was quite large, allowing the analysis of increasing doses of variant alleles and interactions between the 2 genes. The main potential weakness of our study was the relatively crude estimation of NSAID use. We defined NSAID use as 3 or more times per week for 1 month within the past 2 years, similar to definitions used in other epidemiologic studies in which associations with aspirin and colon cancer were detected. 18 Although it certainly can be argued that this is not the optimal definition of NSAID use, the fact that we see similar overall results with respect to the protective effects of ibuprofen and aspirin as those seen in other studies would suggest that this was an accurate measure of the protective effect of these agents. Complete dosage information over a lifetime is difficult to ascertain accurately, but it is possible that our definition identifies the subset of chronic NSAID users. Our lack of complete dosage information could affect the results of the study, most likely diminishing our ability to detect associations with the enzyme genotypes. However, this would not be expected to lead to false-positive results, such as the interactions we observed between variant CYP2C9 genotypes, use of ibuprofen, and decreased risk of colorectal cancer. In summary, the slower-metabolizing variants of CYP2C9 enhanced the chemopreventive activity of ibuprofen against colorectal cancer, and this effect was more pronounced in those who were wild type for UGT1A6. As in most large epidemiologic studies, it is possible that some of our findings are owing to chance and/or multiple comparisons. One argument against this, however, is that the effects of the slower-metabolizing variants we observed were what one would hypothesize and what is reasonable biologically: an intensification of the anticancer effect of ibuprofen resulting from slower catabolism of the active drug by the relevant enzyme, CYP2C9. Also, the dose-response relationship that was present with respect to increasing numbers of variant CYP2C9 alleles is strong evidence that their interaction with ibuprofen was not simply owing to chance. Still, we would hope that other studies would try to reproduce our findings and thereby lend strength to their validity. References 1. Baron JA, Sandler RS. Nonsteroidal anti-inflammatory drugs and cancer prevention. Annu Rev Med 2000;51: Leemann TD, Transon C, Bonnabry P, et al. A major role for cytochrome P450TB (CYP2C subfamily) in the actions of nonsteroidal antiinflammatory drugs. Drugs Exp Clin Res 1993;19: Hutt AJ, Caldwell J, Smith RL. The metabolism of aspirin in man: a population study. Xenobiotica 1986;16: Chan AT, Tranah GJ, Giovannucci EL, et al. Genetic variants in the UGT1A6 enzyme, aspirin use, and the risk of colorectal adenoma. J Natl Cancer Inst 2005;97: Ciotti M, Marrone A, Potter C, et al. Genetic polymorphism in the human UGT1A6 (planar phenol) UDP-glucuronosyltransferase: pharmacological implications. Pharmacogenetics 1997;7: Takanashi K, Tainaka H, Kobayashi K, et al. CYP2C9 Ile359 and Leu359 variants: enzyme kinetic study with seven substrates. Pharmacogenetics 2000;10: Chan AT, Tranah GJ, Giovannucci EL, et al. A prospective study of genetic polymorphisms in the cytochrome P-450 2C9 enzyme and the risk for distal colorectal adenoma. Clin Gastroenterol Hepatol 2004;2: Bigler J, Whitton J, Lampe JW, et al. CYP2C9 and UGT1A6 genotypes modulate the protective effect of aspirin on colon adenoma risk. Cancer Res 2001;61: Crespi CL, Miller VP. The R144C change in the CYP2C9*2 allele alters interaction of the cytochrome P450 with NADPH: cytochrome P450 oxidoreductase. Pharmacogenetics 1997;7: McGreavey LE, Turner F, Smith G, et al. No evidence that polymorphisms in CYP2C8, CYP2C9, UGT1A6, PPARdelta and PPARgamma act as modifiers of the protective effect of regular NSAID use on the risk of colorectal carcinoma. Pharmacogenet Genomics 2005;15: Slattery ML, Samowitz W, Hoffman M, et al. Aspirin, NSAIDs, and colorectal cancer: possible involvement in an insulin-

8 July 2006 CYP2C9, UGT1A6, AND COLORECTAL CANCER 901 related pathway. Cancer Epidemiol Biomarkers Prev 2004;13: Slattery ML, Potter J, Caan B, et al. Energy balance and colon cancer beyond physical activity. Cancer Res 1997;57: Edwards S, Slattery ML, Mori M, et al. Objective system for interviewer performance evaluation for use in epidemiologic studies. Am J Epidemiol 1994;140: Dorado P, Berecz R, Norberto MJ, et al. CYP2C9 genotypes and diclofenac metabolism in Spanish healthy volunteers. Eur J Clin Pharmacol 2003;59: Peters WH, te Morsche RH, Roelofs HM. Combined polymorphisms in UDP-glucuronosyltransferases 1A1 and 1A6: implications for patients with Gilbert s syndrome. J Hepatol 2003;38: Castillo M, Smith PC. Disposition and reactivity of ibuprofen and ibufenac acyl glucuronides in vivo in the rhesus monkey and in vitro with human serum albumin. Drug Metab Dispos 1995;23: Kuehl GE, Lampe JW, Potter JD, et al. Glucuronidation of nonsteroidal anti-inflammatory drugs: identifying the enzymes responsible in human liver microsomes. Drug Metab Dispos 2005;33: Rosenberg L, Palmer JR, Zauber AG, et al. A hypothesis: nonsteroidal anti-inflammatory drugs reduce the incidence of largebowel cancer. J Natl Cancer Inst 1991;83: Address requests for reprints to: Wade S. Samowitz, MD, Department of Pathology, University of Utah Health Sciences Center, Salt Lake City, Utah wade.samowitz@aruplab.com; fax: (801) Supported by grants CA48998 and CA61757 from the National Cancer Institute (to M.L.S.). This research was supported by the Utah Cancer Registry, which is funded by contract N01-PC from the National Cancer Institute, with additional support from the State of Utah Department of Health and the University of Utah, the Northern California Cancer Registry, and the Sacramento Tumor Registry. The contents of this article are solely the responsibility of the authors and do not necessarily represent the official view of the National Cancer Institute. The authors would like to acknowledge the contributions of Dr Bette Caan, Sandra Edwards, Leslie Palmer, and Judy Morse to the data collection and management efforts of this study and of Michael Hoffman and Erica Wolff for genotyping.