Ornithine Decarboxylase Polymorphism Modification of Response to Aspirin Treatment for Colorectal Adenoma Prevention

Transcription

1 Ornithine Decarboxylase Polymorphism Modification of Response to Aspirin Treatment for Colorectal Adenoma Prevention Elizabeth L. R. Barry, John A. Baron, Shubha Bhat, Maria V. Grau, Carol A. Burke, Robert S. Sandler, Dennis J. Ahnen, Robert W. Haile, Thomas G. O Brien Background: Previous research suggests that the G315A singlenucleotide polymorphism in the ornithine decarboxylase (ODC) gene may be a genetic marker for risk of colorectal neoplasia and may also modify the association of aspirin use with risk. Methods: We tested these hypotheses among participants in the Aspirin/Folate Polyp Prevention Study who were randomly assigned to placebo or to aspirin treatment (81 or 325 mg daily) and followed for 3 years for the occurrence of new adenomas. Genomic DNA from 973 subjects was analyzed for. Relative risks (RRs) and 95% confidence intervals (CIs) were calculated to test the association between and adenoma occurrence and interactions with aspirin treatment. All statistical tests were two-sided. Results: Of the 973 subjects, 54% were homozygous wild-type (GG), 7% were homozygous variant (AA), and 39% were heterozygous individuals; the allele frequencies varied statistically significantly by race and ethnicity. Among these subjects, the absolute risk of any adenoma was 45% and the risk of an advanced lesion was 10%. Overall, no association was found between and the occurrence of new adenomas, but genotype did modify the effect of aspirin on adenoma risk. Although aspirin treatment had no protective effect among subjects with a GG genotype, among subjects with at least one A allele, it was associated with statistically significant reduced risks of any adenoma (RR = 0.77, 95% CI = 0.63 to 0.95; P =.02, P interaction =.04) and of advanced lesions (RR = 0.51, 95% CI = 0.29 to 0.90; P =.02, P interaction =.02). Among subjects with at least one A allele, 40.8% who took aspirin versus 52.9% who took placbo developed adenomas; 7.1% versus 14.0% developed advanced lesions. Conclusion: may modify the response to aspirin treatment for colorectal adenoma prevention. [J Natl Cancer Inst 2006;98: ] Polyamines are cell-signaling molecules whose regulation is important for cell growth and development and for wound healing (1 ). Ornithine decarboxylase (ODC) catalyzes the first step in polyamine biosynthesis the decarboxylation of ornithine to putrescine (2 ). Several lines of evidence implicate polyamines and ODC in carcinogenesis. Polyamine levels and ODC activity are increased in many epithelial cancers, including cancers of the skin, prostate, and colon (1 ). Polyamine synthesis and ODC activity appear to be induced, and polyamine catabolism suppressed, by mutations in adenomatous polyposis coli and K-ras, respectively, two genes that are commonly mutated in colon cancer (3 8 ). Conversely, polyamine catabolism appears to be induced by nonsteroidal anti-inflammatory drugs (NSAIDs), whose use is associated with a decreased risk of several epithelial cancers, including colorectal cancer (9 ). In humans, a single-nucleotide polymorphism (SNP) in the ODC gene is thought to reduce its expression (10 ). The polymor- phic site (G315A) is located in a regulatory region in the first intron near the binding sites for transcription factors. Evidence that this ODC polymorphism has functional relevance comes from in vitro studies showing that this region influences the binding of transcription factors (11 ) and studies of cultured cells showing differential transcriptional regulation of the two alleles (10,12 ). A previous genetic association study reported that individuals homozygous for the minor A allele were approximately half as likely to have an adenoma occurrence as GG (wild-type) homozygotes, and the risk of adenoma was even lower in AA homozygotes who reported taking aspirin (10 ). These results suggest that this ODC polymorphism may be a genetic marker for risk of colorectal neoplasia and may also modify the association of aspirin use with risk. The present analysis was undertaken to examine the relationships among this ODC polymorphism, aspirin treatment, and the risk of colorectal adenoma among subjects at high risk of adenoma who were randomly assigned to placebo or aspirin treatment in the Aspirin/Folate Polyp Prevention Study (13 ), which showed that low-dose aspirin has a moderate chemopreventative effect on adenomas in the large bowel. We tested the hypotheses that the ODC AA genotype is associated with a reduction in adenoma occurrence and that modifies the reduction in adenoma risk associated with aspirin treatment. S UBJECTS AND METHODS Study Population The subjects were participants in the Aspirin/Folate Polyp Prevention Study, a double-blinded placebo-controlled clinical trial of aspirin and folate as chemopreventive agents against the occurrence of new colorectal adenomas ( Fig. 1, clinical trial registration no. NCT ) (13 ). All subjects had a recent history of a histologically confirmed colorectal adenoma and a complete colonoscopy within 3 months before enrollment (qualifying examination), with no known polyps left in the bowel. A total of 1121 subjects were randomly assigned to one of three study arms Affiliations of authors: Departments of Community and Family Medicine (ELRB, JAB, MVG) and Medicine (JAB), Dartmouth Medical School, Lebanon, NH; Lankenau Institute for Medical Research, Wynnewood, PA (SB, TGO); Department of Gastroenterology, Cleveland Clinic Foundation, Cleveland, OH (CAB); Department of Medicine, Division of Gastroenterology and Hepatology, University of North Carolina, Chapel Hill, NC (RSS); Department of Medicine, University of Colorado Health Sciences Center, Denver, CO (DJA); Department of Preventive Medicine, University of Southern California, Los Angeles, CA (RWH). Correspondence to: Elizabeth L. R. Barry, PhD, Department of Community and Family Medicine, Dartmouth Medical School, 46 Centerra Parkway, Ste. 300, Lebanon, NH ( elizabeth.l.barry@dartmouth.edu ). See Notes following References. DOI: /jnci/djj398 The Author Published by Oxford University Press. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 20, October 18, 2006

2 History of histologically confirmed colorectaladenoma and complete colonoscopy within 3 months prior to enrollment (n = 1409) Failed to complete run-in (n = 288) Randomized (n = 1121) Fig. 1. Trial flow diagram for the Aspirin/Folate Polyp Prevention Study. Outcomes (adenomas and advanced lesions at follow-up endoscopy exam) are shown for those subjects in each aspirin treatment group that underwent ornithine decarboxylase genotyping (N = 973). Allocated to placebo (n = 372) Allocated to 81 mg aspirin (n = 377) Allocated to 325 mg aspirin (n = 372) between September 25, 1994, and July 15, 1998, at recruitment sites associated with nine clinical centers in North America. Subjects gave written informed consent, and all research (including this analysis) was approved by the research ethics committees of the participating institutions. Subjects were randomly assigned to placebo, 81 mg of aspirin (lower dose), or 325 mg of aspirin (higher dose) daily. Subjects were also independently randomly assigned to placebo or 1 mg of folate daily in a factorial design. At enrollment, subjects agreed to avoid the use of NSAIDs outside the study while they were receiving their randomly assigned treatment; NSAID use was monitored by questionnaires every 4 months. By protocol, follow-up colonoscopies were to be scheduled 3 years after the qualifying examination. A total of 1084 subjects (97%) completed a follow-up colonoscopy during the period from at least 1 year after random assignment through the anticipated 3-year follow-up (mean intervention period ± standard deviation, 32.7 ± 3.8 months) and were considered suitable for inclusion in the present analysis. Of them, 471 (43%) were diagnosed with at least one adenoma. Findings from interim endoscopic examinations, which were performed after random assignment but before the follow-up examination in approximately 5% of the subjects, were included in the analysis. A single study pathologist (D.C.S.) reviewed all clinical samples removed from the large bowel after random assignment and categorized the lesions as nonneoplastic or neoplastic (i.e., adenomatous). Adenomas were characterized as to degree of dysplasia and architecture: tubular, tubulovillous (25% 75% villous component), or villous (>75% villosity). Genotyping Lost to follow-up (n = 5) Death (n = 3) No genotyping (n = 46) Completed follow-up exam (n = 326) 155 with any adenoma 38 with advanced lesion Of the 1084 subjects who completed a follow-up colonoscopy, ODC genotyping was performed on 973 (90%). The remaining 111 subjects (10%) could not be included because of lack of consent. Genomic DNA was isolated from whole blood cells using proteinase K digestion and phenol chloroform extraction. ODC was genotyped using a TaqMan-based allelic discrimination assay as described (12 ). Briefly, a 172-bp fragment containing the polymorphic site (SNP accession no. rs ) was amplified using the polymerase chain reaction, and the SNP was detected with allele-specific probes containing different 5 fluorescent labels. All DNA samples were analyzed in duplicate or triplicate. Of the samples analyzed by the TaqMan assay, 7% were also analyzed by a restriction fragment length polymorphism polymerase chain reaction assay based on the polymorphic Pst1 site (12 ) with identical results. Race and Ethnicity At enrollment, subjects were asked to complete a questionnaire on which they identified their race or ethnic background by selecting one of the following ( Table 1 ): 1) white, not of Hispanic origin; 2) black, not of Hispanic origin; 3) Hispanic; 4) American Indian or Alaskan Native; 5) Asian or Pacific Islander; 6) other; 7) uncertain. For multivariable analysis, a bivariate race/ethnicity variable was created: self-reported non- Hispanic whites (answer 1, N = 860) and all others (answers 2 6 above, N = 113). There were no uncertain responses. Statistical Analysis Lost to follow-up (n = 7) Death (n = 3) No genotyping (n = 49) Completed follow-up exam (n = 328) 126 with any adenoma 25 with advanced lesion Lost to follow-up (n = 13) Death (n = 4) No genotyping (n = 53 ) Completed follow-up exam (n = 319) 153 with any adenoma 36 with advanced lesion Comparisons of subject characteristics by at SNP rs ( Tables 1 and 2 ) were performed using Pearson chi-square tests for categorical variables and analysis of variance or nonparametric Kruskal Wallis tests for continuous variables. Deviations from Hardy Weinberg equilibrium were assessed by the likelihood ratio chi-square test. Journal of the National Cancer Institute, Vol. 98, No. 20, October 18, 2006 ARTICLES 1495

3 Table 1. Race and ethnicity of subjects by ornithine decarboxylase (ODC) genotype * Table 2. Baseline characteristics of subjects by ornithine decarboxylase (ODC) genotype, N (%) GG GA AA GG GA AA Race/ethnicityN (%) P Characteristic P * All 973 (100) 522 (53.6) 383 (39.4) 68 (7.1) White, non-hispanic 860 (88.4) 476 (55.4) 336 (39.1) 48 (5.6) Referent Black, non-hispanic 45 (4.6) 21 (46.7) 17 (37.8) 7 (15.6).02 Hispanic 45 (4.6) 22 (48.9) 16 (35.6) 7 (15.6).02 Asian, Pacific Islander 16 (1.6) 1 (6.25) 10 (62.5) 5 (31.25) <.001 American Indian, 3 (0.3) 1 (33.3) 1 (33.3) 1 (33.3).11 Alaskan native Other 4 (0.4) 1 (25) 3 (75) 0.33 Minorities combined 113 (11.6) 46 (40.7) 47 (41.6) 20 (17.7) <.001 * The study population is 973 participants in the Aspirin/Folate Poly Prevention Study. Pearson chi-square tests (two-sided) for comparisons with the white, non-hispanic category. This group combines all categories except the white, non-hispanic category. The principal outcome of the trial was the occurrence of one or more adenomas during randomized treatment. We also evaluated the occurrence of advanced lesions: tubulovillous adenomas, villous adenomas, large adenomas ( 1 cm in diameter), adenomas with severe dysplasia, and invasive cancer. Risk ratios were calculated using an overdispersed generalized linear model for the Poisson distribution as an approximation to the binomial family. Unadjusted (univariate) relative risks (RRs) are presented because the small numbers in some groups limited the number of covariates appropriate for the models. Where indicated, adjusted relative risks were calculated by including age, sex, race, aspirin treatment group, and mean follow-up time as covariates. Interactions were evaluated using product interaction terms and Wald tests in models for the full data set as well as in those for a caseonly analysis ( 14 ). Both approaches produced essentially identical results (data not shown). All statistical tests were two-sided, and all analyses were performed under the intention-to-treat principle. R ESULTS Gene Frequencies and Subject Characteristics The overall distribution was similar to that seen in previous studies, with an A allele frequency of 0.27: 54% were homozygous wild type (GG), 7% were homozygous variant (AA), and 39% were heterozygous individuals. The genotypes were in Hardy Weinberg equilibrium ( P =.79). However, the gene frequency varied statistically significantly by self-reported race and ethnicity ( Table 1 ). The prevalence of the AA genotype was much lower among non-hispanic whites (approximately 6%) than among minorities combined (approximately 18% overall) ( P <.001). Among minorities, the frequency of AA homozygotes ranged from approximately 16% in Hispanics and Blacks to approximately 31% in Asians. Subject demographic and study characteristics were similar across s ( Table 2 ). No difference in the number of baseline adenomas per subject was observed across s. However, the proportion of subjects with one or more advanced adenomas was lower among subjects with either the GA or AA genotype (25%) than among those with the GG genotype No. of subjects Mean age, y (±SD) 57.9 (±9.4) 57.7 (±9.8) 56.2 (±9.0).41 Male sex, N (%) 333 (64) 253 (66) 43 (63).76 Baseline adenomas Mean number Proportion of subjects with 164 (31) 93 (24) 18 (26).06 advanced adenomas, N (%) Aspirin treatment, N (%).89 Placebo 169 (32) 135 (35) 22 (32) 81 mg aspirin 178 (34) 125 (33) 25 (37) 325 mg aspirin 175 (34) 123 (32) 21 (31) Folate treatment, N (%) 245 (51) 169 (48) 34 (52).53 Mean follow-up time, mo (±SD) 32.7 (±3.4) 32.4 (±3.4) 34.0 (±4.7).01 * Statistical tests (Pearson chi-square tests for categorical variables and analysis of variance or nonparametric Kruskal Wallis tests for continuous variables) for difference between subjects with different s were two-sided. SD = standard deviation. Advanced adenomas include tubulovillous adenomas (25% 75% villous features), villous adenomas (more than 75% villous), large adenomas (at least 1 cm in diameter), or adenomas with severe dysplasia. Some subjects were not randomly assigned to folate treatment (46 participants with the GG, 29 with GA, and 3 with AA). Number of months from the random assignment date to the follow-up colonoscopy date. (31%) ( P =.02). In addition, subjects with the AA genotype had slightly longer (approximately 1.5 months) follow-up times on average than subjects with either the GG or GA genotypes. No statistically significant differences were found in other characteristics (age, sex, treatment assignments) among the subjects when grouped by. Main Association of ODC Genotype The frequency of adenoma occurrence, which was approximately 45% overall, was similar (between 41.2% and 45.7%) among subjects in the three subgroups ( Table 3 ). As indicated by the relative risks, there was virtually no association between the and the risk of any adenoma. However, when the analysis was limited to advanced lesions (found in 10% of subjects), there was a nonstatistically significant 30% reduction in risk associated with the AA genotype relative to the GG genotype (RR = 0.70, 95% confidence interval [CI] = 0.29 to 1.69). Among heterozygotes, there was a nonstatistically significant 10% reduction in risk of advanced lesions compared with the GG genotype. To avoid possible confounding by population stratification, we also conducted analyses limited to non-hispanic whites (88% of subjects). In this population, there were nonstatistically significant risk reductions for advanced lesions among heterozygotes (RR = 0.86, 95% CI = 0.57 to 1.30) and among AA homozygotes (RR = 0.18, 95% CI = 0.03 to 1.20). Interaction Analysis In the subset of 973 subjects who underwent genotyping, aspirin treatment (low and high doses combined) was associated with nonstatistically significant risk reductions for all adenomas (RR = 0.91, 95% CI = 0.78 to 1.05) and for advanced lesions 1496 ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 20, October 18, 2006

4 Table 3. Association of ornithine decarboxylase (ODC) genotype with risk of adenoma occurrence * Occurrence/ total (%) RR (95% CI) Adjusted RR (95% CI) All adenomas GG 231/522 (44.3) GA 175/383 (45.7) 1.03 (0.89 to 1.2) 1.03 (0.89 to 1.20) AA 28/68 (41.2) 0.93 (0.69 to 1.25) 0.98 (0.73 to 1.32) Advanced lesions GG 56/522 (10.7) GA 38/383 (9.9) 0.92 (0.63 to 1.37) 0.90 (0.61 to 1.34) AA 5/68 (7.4) 0.69 (0.29 to 1.63) 0.70 (0.29 to 1.69) * RR = relative risk; CI = confidence interval. Adjusted for age, sex, race, aspirin treatment group, and mean follow-up time. (RR = 0.81, 95% CI = 0.55 to 1.19) ( Table 4 ). However, the relationship between aspirin and risk varied by. Adenoma risk was not associated with aspirin treatment among subjects with the GG genotype, but among subjects with at least one A allele, aspirin treatment was associated with statistically signifi cant risk reductions for all adenomas (RR = 0.77, 95% CI = 0.63 to 0.95; P =.02) and for advanced lesions (RR = 0.51, 95% CI = 0.29 to 0.90; P =.02). Among subjects with at least one A allele, 40.8% who took aspirin versus 52.9% who took placbo developed adenomas; 7.1% versus 14.0% developed advanced lesions. The interaction between and combined aspirin treatment was statistically significant for all adenomas ( P =.04) and for advanced lesions ( P =.02). Essentially identical and statistically significant results were obtained when these analyses were adjusted for potential confounding factors (race, sex, age, and follow-up interval) or restricted to non-hispanic whites only (data not shown). An interaction between and aspirin treatment was also reflected in differences between genotype-specific relative risks among aspirin treatment groups ( Table 5 ). Among subjects in the placebo group, those with at least one A allele had nonstatistically significant higher risks for all adenomas (RR = 1.24, 95% CI = 0.99 to 1.56) and for advanced lesions (RR = 1.48, 95% CI = 0.81 to 2.72) than subjects with the GG genotype. Conversely, among subjects in the aspirin treatment groups, those with at least one A allele had nonstatistically significant lower risks for all adenomas (RR = 0.91, 95% CI = 0.76 to 1.08) and for advanced lesions (RR = 0.63, 95% CI = 0.38 to 1.04) than subjects with the GG genotype. Essentially identical results were obtained when these analyses were adjusted for potential confounding factors or were restricted to non-hispanic whites (data not shown). As previously reported for the trial as a whole, the lower dose of aspirin (81 mg) was more effective than the higher dose (325 mg) at reducing the risk of colorectal adenoma ( 13 ). The pattern was also seen in the subset of subjects included in this analysis ( Table 6 ). The lower dose of aspirin treatment was associated with statistically significant risk reductions for all adenomas (RR = 0.81, 95% CI = 0.68 to 0.96; P =.02) and nonstatistically significant risk reductions for advanced lesions (RR = 0.65, 95% CI = 0.40 to 1.06; P =.08), whereas the higher dose was not associated with risk reductions. In the analysis stratified by genotype, neither the low nor the high dose of aspirin treatment was associated with risk reductions among subjects with the GG genotype. Among heterozygotes, the lower aspirin dose was associated with statistically significant risk reductions for all adenomas (RR = 0.65, 95% CI = 0.49 to 0.85; P =.002) and for advanced lesions (RR = 0.27, 0.11 to 0.69; P =.006), whereas the higher dose showed smaller, nonstatistically significant reductions. Among AA homozygotes, both aspirin doses were also associated with risk reductions for all adenomas and advanced lesions, with more pronounced reductions in the higher dose group. Possibly because of the small numbers of AA subjects, however, the corresponding relative risks were not statistically significantly different. Essentially identical results were obtained when these analyses were restricted to non-hispanic whites (data not shown). D ISCUSSION In this analysis of a randomized aspirin trial, we found that modified the effect of aspirin treatment on the occurrence of new colorectal adenomas. Although aspirin did not reduce adenoma risk among subjects with the ODC GG genotype, aspirin treatment conferred a statistically significant risk reduction among subjects with at least one ODC A allele. We began this analysis with the hypothesis that the ODC AA genotype would be associated with a reduced risk of occurrence of new colorectal adenomas, as reported by Martinez et al. (10 ). In that study, the odds of adenoma occurrence among AA homozygotes was about half of that for heterozygotes and GG homozygotes, a result that reached borderline statistical significance ( P =.05). Although our data provided no evidence for an association between the AA genotype and reduced overall adenoma Table 4. Association of aspirin treatment with risk of adenoma occurrence stratified by ornithine decarboxylase (ODC) genotype All subjects GGGA or AA Treatment Occurrence/total (%) RR (95% CI) * Occurrence/total (%) RR (95% CI) Occurrence/total (%) RR (95% CI) All adenomas Placebo 155/326 (47.6) /169 (42.6) /157 (52.9) 1.0 Aspirin 279/647 (43.1) 0.91 (0.78 to 1.05) 159/353 (45.0%) 1.06 (0.86 to 1.30) 120/294 (40.8) 0.77 (0.63 to 0.95) Advanced lesions Placebo 38/326 (11.7) /169 (9.5) /157 (14.0) 1.0 Aspirin 61/647 (9.4) 0.81 (0.55 to 1.19) 40/353 (11.3) 1.20 (0.69 to 2.07) 21/294 (7.1) 0.51 (0.29 to 0.90) * RR = relative risk; CI = confidence interval. Unadjusted relative risks are shown. Interaction between and aspirin treatment, P =.04 (two-sided Wald test). Combined 81 and 325 mg aspirin treatment groups. Interaction between and aspirin treatment, P =.02 (two-sided Wald test). Journal of the National Cancer Institute, Vol. 98, No. 20, October 18, 2006 ARTICLES 1497

5 Table 5. Association of ornithine decarboxylase (ODC) genotype with risk of adenoma occurrence stratified by aspirin treatment group Aspirin treatment All subjects PlaceboAspirin * Occurrence/total (%) RR (95% CI) Occurrence/total (%) RR (95% CI) Occurrence/total (%) RR (95% CI) All adenomas GG 231/522 (44.3) /169 (42.6) /353 (45.0) 1.0 GA or AA 203/451 (45.0) 1.02 (0.88 to 1.17) 83/157 (52.9) 1.24 (0.99 to 1.56) 120/294 (40.8) 0.91 (0.76 to 1.08) Advanced lesions GG 56/522 (10.7) /169 (9.5) /353 (11.3) 1.0 GA or AA 43/451 (9.5) 0.89 (0.61 to 1.30) 22/157 (14.0) 1.48 (0.81 to 2.72) 21/294 (7.1) 0.63 (0.38 to 1.04) * Combined 81 and 325 mg aspirin treatment groups. RR = relative risk; CI = confidence interval. Unadjusted relative risks are shown. Interaction between and aspirin treatment, P =.04 (two-sided Wald test). Interaction between and aspirin treatment, P =.02 (two-sided Wald test). occurrence, this genotype was associated with a nonstatistically significant 30% reduction in the occurrence of new advanced lesions. Data for advanced adenomas were not reported by Martinez et al. (10 ). Notably, in both our study and that of Martinez et al., the statistical power to detect associations with the AA genotype was similarly low as a consequence of the low prevalence of the ODC A allele. Thus, larger studies will be required to determine whether is a marker of overall risk of adenoma. Interestingly, in the present study, a lower proportion of subjects with at least one A allele than of subjects with two G alleles had advanced adenomas at baseline (25% versus 31%, respectively, P =.02). These data provide additional support for a relationship between and the occurrence of advanced adenomas. A second hypothesis, also based on the Martinez et al. study (10 ), is that there is an interaction between and the effect of aspirin on the occurrence of new adenomas. In that study, the association of the AA genotype with a decreased risk of adenoma recurrence appeared to be stronger among aspirin users than among nonusers. However, in the present study, the association between the ODC A allele and the risk of adenoma occurrence appeared to be opposite among subjects treated with placebo (higher risk) compared with subjects treated with aspirin (lower risk). In the current study, the protective effect of aspirin treatment on risk of adenoma was confined to subjects with at least one A allele. However, when heterozygotes and AA homozygotes were analyzed separately, the numbers in the AA group were too small to rule out chance. In addition, we found no clear gene dose response for the beneficial effect of the ODC A allele on the association of aspirin treatment with reduced risk of adenoma. Although this apparent lack of a dose response may be due to the limited power and wide confidence intervals in the AA group, it is also possible that a single A allele is sufficient for the biologic effect (e.g., responsiveness to aspirin). The mechanism by which may influence the effectiveness of aspirin treatment in adenoma prevention is not known. Martinez et al. ( 10 ) suggested that the ODC polymorphism and aspirin may each act independently of the other to reduce adenoma risk by suppressing synthesis and activating catabolism, respectively, of colonic muscosal polyamines. This hypothesis is based on their experiments using a human colon cancer cell line, in which the activity of the ODC A allele promoter was selectively suppressed by expression of the wild-type adenomatous polyposis coli gene and by the transcriptional regulator mitotic arrest deficient 1. In addition, although there was no effect of aspirin on ODC expression or activity, aspirin treatment induced the expression and activity of spermidine spermine Table 6. Association of low- and high-dose aspirin treatment with risk of adenoma occurrence stratified by ornithine decarboxylase (ODC) genotype All subjectsgggaaa Treatment N (%)* RR (95% CI) N (%) RR (95% CI) N (%) RR (95% CI) N (%) RR (95% CI) All adenomas Placebo 155/326 (47.6) 1.0 Aspirin (81 mg) 126/328 (38.4) 0.81 (0.68 to 0.96) Aspirin (325 mg) 153/319 (48.0) 1.01 (0.85 to 1.19) Advanced lesions Placebo 38/326 (11.7) 1.0 Aspirin (81 mg) 25/328 (7.6) 0.65 (0.40 to 1.06) Aspirin (325 mg) 36/319 (11.3) 0.97 (0.63 to 1.49) 72/169 (42.6) /178 (40.5) 0.95 (0.74 to 1.21) 87/175 (49.7) 1.17 (0.92 to 1.47) 16/169 (9.5) /178 (10.1) 1.07 (0.56 to 2.02) 22/175 (15.6) 1.33 (0.72 to 2.44) 72/135 (53.3) /125 (34.4) 0.65 (0.49 to 0.85) 60/123 (48.8) 0.91 (0.71 to 1.18) 20/135 (14.8) 1.0 5/125 (4) 0.27 (0.11 to 0.69) 13/123 (10.6) 0.71 (0.37 to 1.39) 11/22 (50) /25 (44.0) 0.88 (0.46 to 1.70) 6/21 (28.6) 0.57 (0.26 to 1.25) 2/22 (9.1) 1.0 2/25 (8) 0.88 (0.13 to 6.06) 1/21 (4.8) 0.52 (0.05 to 5.57) * No. of subjects with an adenoma or advanced lesion divided by the total number of subjects in that group. RR = relative risk; CI = confidence interval. Unadjusted relative risks are shown ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 20, October 18, 2006

6 N-acetyltransferase, an enzyme that is involved in polyamine catabolism. It is possible that the combination of these two effects, via the ODC polymorphism and aspirin, may reduce the risk of colorectal neoplasia to a much greater degree than either alone. Alternatively, or in addition, aspirin may act directly on ODC itself or on a regulator of ODC expression or activity. In vivo studies on the mechanism underlying these observations are needed. In the current study, the overall frequency of the s observed was similar to that seen in other studies involving US populations (10,12,15 ). However, as reported previously (12,16 ), there were statistically significant differences in ODC allele frequencies among different races and ethnic groups. In the present study, the prevalence of the A allele was the lowest in American Caucasians (0.25), a proportion similar to that reported previously (12,16 ). However, the prevalence of the A allele in American Hispanics (0.33) was much higher than previously reported in non-american Hispanics (0.18) (12,16 ). The prevalence of the A allele was high in African Americans (0.34), in agreement with a previous report (12,16 ), and was highest in American Asians (0.63), although this result was based on a very small sample size. ODC genotyping of a large number of Japanese and Chinese Asians supports these data (O Brien TG: unpublished observations). This variability in genotype frequencies suggests the potential for confounding of our results due to population stratification if race or ethnicity is also associated with a substantial difference in disease frequency. Although there are little data on the risk of adenoma recurrence by race and ethnicity, the incidence of colorectal cancer does vary by race and ethnicity (17 ). However, because our results did not change when the analyses were restricted to non-hispanic whites, it is unlikely that they are due to confounding by population stratification. There were some limitations to the current analyses. We had limited power to study advanced lesions due to the small number of these lesions in subjects under regular surveillance. In addition, as discussed above, the low number of AA homozygotes limited our ability to study associations in these subjects. However, the study also had several strengths. An important advantage of our study is that aspirin treatment was randomly assigned to the participants during the trial, whereas in the study of Martinez et al. (10 ), aspirin use was assessed only by a baseline questionnaire asking whether aspirin had been used in the previous month. Another strength of the current study is that a single pathologist reviewed all lesions from the subjects. There are suggestions that variants in other genes may also modify the effect of aspirin or NSAIDs on risk of colorectal neoplasia. These genes include those that encode enzymes involved in NSAID metabolism, such as uridine diphosphate glucuronosyltransferase 1A6 and cytochrome P450 2C9 ( 18, 19 ), and prostaglandin synthesis and metabolism, such as cyclooxygenases-1 and -2, prostacyclin synthase, and arachidonate 5-lipoxygenase ( 20, 21, 22 ). However, there is weak or conflicting evidence for the involvement of variants in these genes in colorectal neoplasia ( 23 ). In conclusion, our results do not provide support for the hypothesis that the G315A ODC polymorphism is a genetic marker for colorectal adenoma occurrence, although we did find suggestions of an association with advanced adenomas. Our findings do suggest that may be an important predictor of the response to aspirin use for adenoma chemoprevention. Future studies should examine the association between and colorectal cancer (rather than adenoma occurrence). Due to the very small numbers of ODC AA homozygotes, larger numbers of subjects would be required to address these hypotheses with sufficient power. R EFERENCES (1) Gerner EW, Meyskens FL Jr. Polyamines and cancer: old molecules, new understanding. Nat Rev Cancer 2004 ;4 : (2) Pegg AE, Shantz LM, Coleman CS. Ornithine decarboxylase: structure, function and translational regulation. Biochem Soc Trans 1994 ; 22 : (3) Giardiello FM, Hamilton SR, Hylind LM, Yang VW, Tamez P, Casero RA Jr. Ornithine decarboxylase and polyamines in familial adenomatous polyposis. Cancer Res 1997 ; 57 : (4) He TC, Sparks AB, Rago C, Hermeking H, Zawel L, da Costa LT, et al. Identification of c-myc as a target of the APC pathway. Science 1998 ; 281 : (5) Bello-Fernandez C, Packham G, Cleveland JL. The ornithine decarboxylase gene is a transcriptional target of c-myc. Proc Natl Acad Sci U S A 1993 ;90 : (6) Fultz KE, Gerner EW. APC-dependent regulation of ornithine decarboxylase in human colon tumor cells. Mol Carcinog 2002 ;34 :10 8. (7) Erdman SH, Ignatenko NA, Powell MB, Blohm-Mangone KA, Holubec H, Guillen-Rodriguez JM, et al. APC-dependent changes in expression of genes influencing polyamine metabolism, and consequences for gastrointestinal carcinogenesis, in the Min mouse. Carcinogenesis 1999 ;20 : (8) Ignatenko NA, Babbar N, Mehta D, Casero RA Jr, Gerner EW. Suppression of polyamine catabolism by activated Ki-ras in human colon cancer cells. Mol Carcinog 2004 ;39 : (9) Babbar N, Ignatenko NA, Casero RA Jr, Gerner EW. Cyclooxygenaseindependent induction of apoptosis by sulindac sulfone is mediated by polyamines in colon cancer. J Biol Chem 2003 ;278 : (10) Martinez ME, O Brien TG, Fultz KE, Babbar N, Yerushalmi H, Qu N, et al. Pronounced reduction in adenoma recurrence associated with aspirin use and a polymorphism in the ornithine decarboxylase gene. Proc Natl Acad Sci U S A 2003 ;100 : (11) Walhout AJ, van der Vliet PC, Timmers HT. Sequences flanking the E-box contribute to cooperative binding by c-myc/max heterodimers to adjacent binding sites. Biochim Biophys Acta 1998 ;1397 : (12) Guo Y, Harris RB, Rosson D, Boorman D, O Brien TG. Functional analysis of human ornithine decarboxylase alleles. Cancer Res 2000 ;60 : (13) Baron JA, Cole BF, Sandler RS, Haile RW, Ahnen D, Bresalier R, et al. A randomized trial of aspirin to prevent colorectal adenomas. N Engl J Med 2003 ;348 : (14) Khoury MJ, Flanders WD. Nontraditional epidemiologic approaches in the analysis of gene-environment interaction: case-control studies with no controls! Am J Epidemiol 1996 ; 144 : (15) Visvanathan K, Helzlsouer KJ, Boorman DW, Strickland PT, Hoffman SC, Comstock GW, et al. Association among an ornithine decarboxylase polymorphism, androgen receptor gene (CAG) repeat length and prostate cancer risk. J Urol 2004 ;171 : (16) O Brien TG, Guo Y, Visvanathan K, Sciulli J, McLaine M, Helzlsouer KJ, et al. Differences in ornithine decarboxylase and androgen receptor allele frequencies among ethnic groups. Mol Carcinog 2004 ;41 : (17) Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor A, et al. Cancer statistics, CA Cancer J Clin 2005 ;55 : (18) Bigler J, Whitton J, Lampe JW, Fosdick L, Bostick RM, Potter JD. CYP2C9 and UGT1A6 genotypes modulate the protective effect of aspirin on colon adenoma risk. Cancer Res 2001 ;61 : (19) Chan AT, Tranah GJ, Giovannucci EL, Hunter DJ, Fuchs CS. Genetic variants in the UGT1A6 enzyme, aspirin use, and the risk of colorectal adenoma. J Natl Cancer Inst 2005 ;97 : (20) Ulrich CM, Bigler J, Sparks R, Whitton J, Sibert JG, Goode EL, et al. Polymorphisms in PTGS1 (=COX-1) and risk of colorectal polyps. Cancer Epidemiol Biomarkers Prev 2004 ;13 : (21) Ulrich CM, Whitton J, Yu JH, Sibert J, Sparks R, Potter JD, et al. PTGS2 (COX-2) 765G > C promoter variant reduces risk of colorectal adenoma Journal of the National Cancer Institute, Vol. 98, No. 20, October 18, 2006 ARTICLES 1499

7 among nonusers of nonsteroidal anti-inflammatory drugs. Cancer Epidemiol Biomarkers Prev 2005 ;14 : (22) Poole EM, Bigler J, Whitton J, Sibert JG, Potter JD, Ulrich CM. Prostacyclin synthase and arachidonate 5-lipoxygenase polymorphisms and risk of colorectal polyps. Cancer Epidemiol Biomarkers Prev 2006 ;15 : (23) McGreavey LE, Turner F, Smith G, Boylan K, Timothy Bishop D, Forman D, 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 : N OTES The authors acknowledge the dedication of the participants in the Aspirin/ Folate Polyp Prevention Study. This work was supported by a National Cancer Institute grant R01-CA59005 (J. A. Baron) and funds from the Lankenau Hospital Foundation (T. G. O Brien). The funding agencies had no role in the study design, data collection and analysis, interpretation of the results, or the preparation of the manuscript. Manuscript received March 10, 2006 ; revised August 1, 2006 ; accepted August 29, ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 20, October 18, 2006