Carcinogenesis vol.28 no.6 pp.1197 1201, 2007 doi:10.1093/carcin/bgl242 Advance Access publication December 6, 2006 Associations of functional polymorphisms in cyclooxygenase-2 and platelet 12-lipoxygenase with risk of occurrence and advanced disease status of colorectal cancer Wen Tan 1,y, Jianxiong Wu 2,y, Xuemei Zhang 1,y, Yongli Guo 1, Junniao Liu 1, Tong Sun 1, Bailin Zhang 2, Dan Zhao 1, Ming Yang 1, Dianke Yu 1 and Dongxin Lin 1, 1 Department of Etiology and Carcinogenesis and 2 Department of Abdominal Surgery, Cancer Institute and Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, 100021, China To whom correspondence should be addressed. Tel: þ86 10 67722460; Fax: þ86 10 67722460; Email: dlin@public.bta.net.cn Aberrant arachidonic acid metabolism by cyclooxygenase (COX)- 2 and 12-lipoxygenase (LOX) has implicated in carcinogenesis. Genetic polymorphisms in COX-2 and 12-LOX might therefore affect susceptibility to colorectal cancer (CRC). To examine this hypothesis, genotypes of COX-2 1290A.G, 1195G.A, 765G.C and 12-LOX 261Arg.Gln polymorphisms were determined in 1000 CRC patients and 1300 controls. Increased risk of developing CRC was associated with the COX-2 1195GA [adjusted odds ratio (OR) 5 1.24, 95% confidence interval (CI) 5 1.00 1.54] and 1195AA (adjusted OR 5 1.77, 95% CI 5 1.38 2.25) genotypes compared with the 1195GG genotype. Similarly, the increased risk for CRC was also associated with the COX-2 765GC genotype (adjusted OR 5 1.73, 95% CI 5 1.23 2.43) compared with the 765GG genotype. Consistent with the results of genotype analyses, the ORs for the A_ 1195 -C_ 765 -containing haplotypes were significantly higher than those for the G_ 1195 -G_ 765 -containing haplotypes (P, 0.01). Furthermore, the 1195A allele was further associated with advanced CRC, with adjusted ORs of Dukes D CRC against Dukes A CRC being 2.43 (95% CI 5 1.15 4.97) and 2.66 (95% CI 5 1.23 5.74) for the 1195GA and 1195AA genotypes, respectively. The increased risk of CRC was also associated with the 12-LOX 261Gln/Gln genotype compared with the Arg/Arg genotype (adjusted OR 5 1.38, 95% CI 5 1.09 1.74). Together, these observations indicate that inherited polymorphisms in arachidonic acid-metabolizing enzymes may confer susceptibility to CRC. Introduction Cyclooxygenases (COXs, also known as prostaglandin endoperoxide H synthases) and arachidonate lipoxygenases (LOXs) are key enzymes in conversion of free arachidonic acid to prostaglandins and hydroxyeicosatetraenoic acids (HETEs) and these resultants are important regulators of many biological processes such as inflammation, immune function, cell proliferation and angiogenesis, which are all relevant to cancer development and progression (1 3). The COX family consists of two isozymes. COX-1 is constitutively expressed and involved in the homeostasis of various physiologic functions, whereas COX-2 is an inducible form and its expression can be induced by pro-inflammatory and mitogenic stimuli (4). It has been documented that both COX-2 mrna and COX-2 protein are up-regulated in colorectal cancer (CRC) and pre-malignant adenomatous colon polyps (5,6). Patients treated with COX inhibitors, such as non-steroidal anti-inflammatory drugs (NSAIDs), had significant reduction in Abbreviations: CI, confidence interval; COX, cyclooxygenase; CRC, colorectal cancer; HETE, hydroxyeicosatetraenoic acid; LOX, lipoxygenase; NSAID, non-steroidal anti-inflammatory drug; OR, odds ratio; SNP, single-nucleotide polymorphism. y These authors contribute equally to this work. CRC risk (7,8), indicating that COX may play an important role in the development of CRC. Human LOX family consists of 5-, 8-, 12- and 15-LOX regarding their positional specificity of arachidonic acid oxygenation to form various HETEs (9,10). Platelet-type 12-LOX, among them, converts arachidonic acid to 12-HETE, and this pathway has also been demonstrated to contribute substantially to the development of cancer including CRC (11 13). The genes encoding COX-2 and 12-LOX are shown to be genetically polymorphic, which may affect the expression or activity of these enzymes and consequently contribute to variation in individual susceptibility to cancer through aberrant arachidonic acid metabolism. We have recently identified several single-nucleotide polymorphisms (SNPs) in the promoter region of the COX-2 gene and demonstrated that the 1195G.A (dbsnp ID: rs689466) change is associated with heightened COX-2 transcription by creating a transcriptional factor c-myb-binding site and elevated risk for developing esophageal squamous cell carcinoma in a Chinese population (14). In our previous study, the 765G.C (dbsnp ID: rs20417) change is also associated with elevated risk of esophageal cancer (14). However, reports regarding the functional consequence of this SNP in literature are contrary (15 17). SNPs have also been reported in the 12-LOX gene, one of which is located in exon 6 (815G.A; dbsnp ID: rs1126667) resulting in an Arg to Gln substitution at amino acid 261 of 12-LOX (18). This amino acid substitution polymorphism is in a highly conserved region of the LOX domain. We have recently characterized the functional significance of 12-LOX 815G.A polymorphism and found that the AA (261Gln/Gln) homozygotes had significantly higher platelet 12- LOX activity and increased risk of developing esophageal squamous cell carcinoma than the GG (261Arg/Arg) homozygotes (19). Because a large body of evidence has demonstrated a strong association between enhanced expression or activity of COX-2 and 12-LOX and CRC (see above), genetic variations that alter gene expression and/or protein activity would anticipate to influence individuals susceptibility to colorectal tumorigenesis and cancer progression. In this study, we performed a large case control study in a Chinese population to evaluate the associations of aforementioned COX-2 and 12-LOX variants, alone and in combination, with the risk of developing CRC. Furthermore, we also analyzed the relationship between these genetic variants and metastatic disease of CRC. Materials and methods Study subjects and clinical characteristics This study included 1000 patients with CRC and 1300 controls. All subjects were ethnic Han Chinese and from Beijing city and its surrounding region. Patients were consecutively recruited between October 1999 and July 2004, at the Cancer Hospital of Chinese Academy of Medical Sciences (Beijing). All patients were newly diagnosed incident cases and histopathologically confirmed as colorectal adenocarcinoma. The response rate for patients was 92%. The exclusion criteria included previous cancer and previous chemotherapy or radiotherapy. Of the eligible patients, 98% underwent coloproctectomy but the rest (2%) did not due to various reasons such as financial difficulties or other diseases of the patients. The diagnosis of patients who did not undergo coloproctectomy was determined by biopsy via endoscopy. The pathological stage of CRC at the time of diagnosis was classified into Dukes A (confined to mucosa), B (varies by system), C (positive lymph nodes) and D (distant metastases). Tumor grade was classified into low (well differentiated), intermediate (moderately differentiated) and high (poorly differentiated) according to the World Health Organization grade classification (20). Of the patients, 29 (2.9%) had family history of CRC. Controls were cancer-free individuals selected from a community cancer screening program for early detection of cancer based on a physical examination, including chest radiography, endoscopy and abdominal ultrasonography, conducted in the same region during the same period when the cases were collected. The participation response rate for Ó The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org 1197
W.Tan et al. controls was 89%. Controls had no individual history of cancer and were frequency matched to patients based on sex and age (±5 years). At recruitment, informed consent was obtained from each subject and each participant was interviewed to collect information on demographic characteristics. This study was approved by the Institutional Review Board of the Chinese Academy of Medical Sciences Cancer Institute. Genotyping of COX-2 and 12-LOX polymorphisms Genomic DNAs were isolated from peripheral blood lymphocytes of the subjects. Genotypes of the COX-2 765G.C, 1195G.A and 1290A.G and 12-LOX 261Arg.Gln polymorphisms were determined by polymerase chain reaction-based restriction fragment length polymorphism methods as described in our previous studies (14,19). The genotypes identified by polymerase chain reaction-based restriction fragment length polymorphism were verified by direct DNA sequencing of randomly selected polymerase chain reaction products of 3% samples. Genotyping was performed without knowledge of subjects patient or control status and a 10% masked, random sample of patients and controls was examined twice by different investigators and the reproducibility was 100%. Haplotype construction and statistical analysis The associations between COX-2 and 12-LOX genotypes and risk of developing CRC were estimated by odds ratios (ORs) and their 95% confidence intervals (CIs) computed by logistic regression. The ORs were adjusted for age and sex. The differences in genotype distributions among subgroups were tested by v 2 test. A P value,0.05 was used as the criterion of statistical significance, and all tests were two sided. All statistical analyses were carried out using Statistical Analysis System software (version 6.12; SAS Institute, Cary, NC). Haplo.score approach was used to test the association of statistically inferred haplotypes with CRC (21). Since haplo.score does not provide the magnitude of the effect of each haplotype, haplo.glm was performed to calculate adjusted ORs and 95% CIs for each haplotype (22). Both haplo.score and haplo.glm were implemented in the haplo.stats software developed using the R language. Results The distributions of mean age and sex did not significantly differ between patients and controls (Table I). The mean age (±SD) was 57.7 (±12.7) years for patients and 57.2 (±9.6) years for controls (P 5 0.31). Males were 59.7% among patients compared with 62.8% among controls (P 5 0.13). Of the patients, 403 had colon cancer and 597 had rectal cancer. Regarding tumor stage, 113, 371, Table I. Distributions of select characteristics among CRC patients and controls Variable Patients, no. (%) Controls, no. (%) P value a Total number 1000 (100) 1300 (100) Sex Male 597 (59.7) 816 (62.8) 0.13 Female 403 (40.3) 484 (37.2) Age (year) 50 301 (30.1) 361 (27.8) 0.29 51 60 222 (22.2) 331 (25.5) 61 70 314 (31.4) 406 (31.2).70 163 (16.3) 202 (15.5) Site of tumor Colon cancer 403 (40.3) Rectal cancer 597 (59.7) Dukes stage at diagnosis A 113 (11.3) B 371 (37.1) C 339 (33.9) D 157 (15.7) Unknown 20 (2.0) Tumor grade Low 129 (12.9) Intermediate 682 (68.2) High 117 (11.7) Unknown 72 (7.2) a Two-sided v 2 test. 1198 339 and 157 patients were classified into Dukes A, B, C and D stages at the time of diagnosis, respectively, whereas 20 patients had unknown stage. For histological differentiation, 129, 682 and 117 patients were classified into low, intermediate and high grade, respectively, whereas 72 patients had missing data. The genotyping results of COX-2 and 12-LOX are shown in Table II. The COX-2 765C, COX-2 1195A, COX-2 1290G and 12-LOX 261Gln allele frequencies were 0.04, 0.57, 0.04 and 0.53 in patients compared with 0.02, 0.50, 0.05 and 0.49 in controls, respectively. The observed genotype frequencies of all four SNPs in controls and patients conformed to the Hardy Weinberg equilibrium. We then analyzed the differences between patients and controls in the distribution of these genotypes and their association with risk of developing CRC. For the 765G.C polymorphism, no 765CC homozygote was found in both patients and controls; however, 765GC heterozygotes were 8.1% in patients compared with 4.8% in controls (adjusted OR 5 1.73, 95% CI 5 1.23 2.43; P, 0.001). Genotype distribution of the 1195G.A polymorphism also significantly differed in patients and controls, with the 1195AA genotype being more prevalent in patients than in controls (32.0 versus 23.7%; P, 0.0001). The adjusted ORs for the 1195GA and 1195AA genotypes in CRC group were 1.24 (95% CI 5 1.00 1.54; P, 0.01) and 1.77 (95% CI 5 1.38 2.25; P, 0.0001), respectively (P trend, 0.001). However, genotype distribution of the 1290A.G polymorphism in patients and controls did not significantly differ and thus this polymorphism was not associated with risk of the cancers. The frequencies of 12-LOX genotypes in patients and controls were also significantly different, with the 261Gln/Gln genotype being over-represented in patients compared with controls (28.7 versus 23.7%; adjusted OR 5 1.38, 95% CI 5 1.09 1.74; P, 0.01), whereas the heterozygous 261Arg/Gln genotype was not significantly different between patients and controls. Similar results were obtained when colon cancer and rectal cancer were separated for analysis (Table II). Linkage disequilibrium analysis showed a D of 0.709 for 1195G.A and 765G.C (P, 0.001), 0.797 for 1290A.G and 1195G.A (P, 0.001) and 0.626 for 1290A.G and 765G.C (P, 0.001) in our study population, which is similar to our previous reports (14,19). We further examined the associations between haplotypes of these SNPs and the risk of CRC. A significant difference in haplotype frequencies was observed between patients and controls (v 2 5 49.98, P, 0.0001, df 5 7). The most prevalent haplotype was A_ 1290 -G_ 1195 -G_ 765 in controls, but A_ 1290 -A_ 1195 -G_ 765 in patients. It was shown that all the 1195A allele-containing haplotypes, except the G_ 1290 -A_ 1195 -G_ 765, were associated with increased risk for developing CRC, with the ORs of 1.32 (95% CI 5 1.17 1.50; P, 0.0001), 2.18 (95% CI 5 1.33 3.59; P, 0.01) and 4.39 (95% CI 5 1.88 10.28; P, 0.0001) for the A_ 1290 -A_ 1195 -G_ 765, G_ 1290 - A_ 1195 -C_ 765 and A_ 1290 -A_ 1195 -C_ 765 haplotypes, respectively. In addition, the ORs for the A_ 1195 -C_ 765 -containing haplotypes were significantly higher than the ORs for the A_ 1195 -G_ 765 -containing haplotype (P, 0.001 or P, 0.01, test for homogeneity), suggesting an interaction between the 1195A and 765C alleles in the context of haplotypes. There were three haplotypes with the frequencies,0.5% in both patients and controls and they were not included in the estimation of the risk because of rarity (Table III). Because both 12-LOX and COX-2 are involved in the arachidonic acid metabolism, whether an interaction exists between these functional SNPs in COX-2 and 12-LOX was tested (Table IV). Although the OR for the presence of both risk genotypes of COX-2 1195AA and 12-LOX 261Gln/Gln (3.43; 95% CI 5 2.05 5.77) was slightly larger than sum minus 1 of the ORs for subjects carrying only one risk genotype (i.e. 2.58 þ 1.68 1 5 3.26), the difference was not significant. However, the results on the multiplicative scale are interesting. The expected OR for the presence of an interaction is 4.33 (i.e. 2.58 1.68) but the observed OR of 3.43 was significantly lower than expected (P, 0.05, test for homogeneity). These findings suggested that subjects carrying the hazard alleles at both loci, thereby having increased levels of both COX-2 and 12-LOX, compared with those carrying the hazard allele at one locus, were at a lower risk for the
Associations of COX-2 and 12-LOX polymorphisms with colorectal cancer Table II. Genotype frequencies of COX-2 and 12-LOX among patients and controls and their association with CRC risk Genotype Controls (n 5 1300) Total patients (n 5 1000) Colon cancer (n 5 403) Rectal cancer (n 5 597) No. (%) No. (%) OR (95% CI) a No. (%) OR (95% CI) a No. (%) OR (95% CI) a COX-2 765G.C GG 1237 (95.2) 919 (91.9) 1.00 (reference) 369 (91.6) 1.00 (reference) 550 (92.1) 1.00 (reference) GC 63 (4.8) 81 (8.1) 1.73 (1.23 2.43) 34 (8.4) 1.78 (1.10 2.70) 47 (7.9) 1.70 (1.15 2.57) CC 0 (0.0) 0 (0.0) NC 0 (0.0) NC 0 (0.0) NC COX-2 1195G.A GG 300 (23.1) 178 (17.8) 1.00 (reference) 65 (16.1) 1.00 (reference) 113 (18.9) 1.00 (reference) GA 692 (53.2) 502 (50.2) 1.24 (1.00 1.54) 199 (49.4) 1.35 (1.00 1.86) 303 (50.8) 1.17 (0.90 1.58) AA 308 (23.7) 320 (32.0) 1.77 (1.38 2.25) 139 (34.5) 2.11 (1.47 2.94) 181 (30.3) 1.61 (1.19 2.12) COX-2 1290A.G AA 1180 (90.8) 914 (91.4) 1.00 (reference) 368 (91.3) 1.00 (reference) 546 (91.4) 1.00 (reference) AG 118 (9.0) 84 (8.4) 0.92 (0.69 1.24) 34 (8.4) 0.95 (0.64 1.43) 50 (8.4) 0.93 (0.62 1.31) GG 2 (0.2) 2 (0.2) 1.35 (0.19 9.67) 1 (0.3) 1.48 (0.28 6.64) 1 (0.2) 1.09 (0.21 5.88) 12-LOX 261Arg.Gln Arg/Arg 329 (25.3) 222 (22.2) 1.00 (reference) 85 (21.1) 1.00 (reference) 137 (22.9) 1.00 (reference) Arg/Gln 663 (51.0) 491 (49.1) 1.10 (0.89 1.35) 202 (50.1) 1.19 (0.89 1.57) 289 (48.4) 1.04 (0.84 1.35) Gln/Gln 308 (23.7) 287 (28.7) 1.38 (1.09 1.74) 116 (28.8) 1.45 (1.06 2.05) 171 (28.7) 1.34 (1.02 1.75) a OR adjusted with sex and age. Table III. Haplotype frequencies of the COX-2 promoter in patients and controls and their association with CRC risk COX-2 promoter haplotype Controls (n 5 1300) Patients (n 5 1000) OR (95% CI) a P value No. of chromosomes (%) No. of chromosomes (%) A_ 1290 -G_ 1195 -G_ 765 1269 (48.8) 848 (42.4) 1.00 (reference) A_ 1290 -A_ 1195 -G_ 765 1191 (45.8) 1032 (51.6) 1.32 (1.17 1.50),0.0001 A_ 1290 -A_ 1195 -C_ 765 10 (0.4) 28 (1.4) 4.39 (1.88 10.28) b,0.0001 G_ 1290 -A_ 1195 -G_ 765 70 (2.7) 36 (1.8) 0.77 (0.51 1.19) 0.20 G_ 1290 -A_ 1195 -C_ 765 37 (1.4) 48 (2.4) 2.18 (1.33 3.59) c,0.01 A_ 1290 -G_ 1195 -C_ 765 8 (0.3) 4 (0.2) NC NC G_ 1290 -G_ 1195 -G_ 765 5 (0.2) 4 (0.2) NC NC G_ 1290 -G_ 1195 -C_ 765 10 (0.4) 0 (0.0) NC NC a Adjusted for sex and age. b P, 0.001, test for homogeneity between A_ 1290 A_ 1195 C_ 765 and A_ 1290 A_ 1195 G_ 765. c P, 0.01, test for homogeneity between G_ 1290 A_ 1195 C_ 765 and G_ 1290 A_ 1195 G_ 765. development of CRC. A similar trend was also observed between COX- 2 765G.C and 12-LOX 261Arg.Gln polymorphisms (Table IV). The associations between COX-2 and 12-LOX genotypes and tumor stage or grade at the time of diagnosis were further evaluated (Table V). No significant association between 12-LOX genotypes and disease status of CRC was observed (data not shown). However, Dukes stage of the cancer at the time of diagnosis was associated with the COX-2 1195G.A polymorphism. The allelic frequencies between the Dukes A patients and patients with more advanced stages, i.e. Dukes B D, were significantly different (v 2 5 8.83, P, 0.05, df 5 3). Logistic regression analysis showed that patients with at least one COX-2 1195A allele (GA or AA genotype) had significantly increased risk for developing more advanced stages of CRC, with the ORs being 1.69 (95% CI 5 1.00 2.89) for Dukes B versus A, 1.74 (95% CI 5 1.02 3.00) for Dukes C versus A and 2.47 (95% CI 5 1.25 4.89) for Dukes D versus A (P trend 5 0.01). Concerning the tumor grade, no significant differences in the allele and genotype frequencies were found among low-, intermediate- and high-grade CRC. The other two COX-2 polymorphisms were either not associated with disease status of the cancer ( 1290A.G polymorphism) or not evaluated because of small number in analysis due to rarity of the polymorphism ( 765G.C). Discussion Because the downstream products of COX-2 and 12-LOX play important roles in colorectal carcinogenesis, we examined whether functional genetic polymorphisms, which enhance the expression or activity of these enzymes, were associated with susceptibility to CRC. On the basis of genotyping 1000 CRC patients (403 colon cancer and 597 rectal cancer) and 1300 controls in Han Chinese population, we found that increased risk for the development of CRC was associated with the COX-2 765C, COX-2 1195A and 12-LOX 261Gln alleles. In addition, we found a significant interaction between the 1195A and 765C alleles in the context of haplotypes in intensifying risk of CRC. These results are consistent with the findings in our previous studies on esophageal cancer (14,19), indicating that functional variants of COX-2 and 12-LOX may confer susceptibility to cancers wherein they play a role. Furthermore, we demonstrated for the first time that the COX-2 1195A allele was also associated with CRC metastasis and patients with this variant allele were at higher risk for the development of advanced cancer compared with the COX-2 1195G allele. A few case control studies have been reported so far with regard to different COX-2 polymorphisms and susceptibility to CRC in different ethnic populations. An amino acid substitution polymorphism in COX-2, 511Val.Ala, which seems to occur only in African-Americans and Asian-Indians, was shown to be associated with reduced risk of colorectal neoplasia in African-Americans (23,24) although inconsistent result also exists (25). Recently, a polymorphism in the untranslated region of exon 10 of COX-2 has been associated with increased risk of CRC in a Spanish population; however, the biological function of this SNP remains unclear (26). At least two studies have been 1199
W.Tan et al. Table IV. Risk of CRC associated with COX-2 genotypes by 12-LOX genotypes COX-2 genotype 12-LOX genotype Arg/Arg a OR (95% CI) b Arg/Gln a OR (95% CI) b Gln/Gln a OR (95% CI) b 1195G.A GG 36/88 1.00 (reference) 93/140 1.61 (0.99 2.68) 49/72 1.68 (0.96 2.91) GA 117/176 1.62 (1.02 2.65) 248/352 1.73 (1.10 2.69) 137/164 2.06 (1.26 3.30) AA 69/65 2.58 (1.51 4.50) 150/171 2.15 (1.32 3.45) 101/72 3.43 (2.05 5.77) 765G.C GG 202/314 1.00 (reference) 449/629 1.08 (0.84 1.40) 268/294 1.40 (1.07 1.86) GC 20/15 2.04 (1.00 4.35) 42/34 1.90 (1.13 3.25) 19/14 2.10 (1.01 4.62) CC 0/0 NC 0/0 NC 0/0 NC a Number of patients/number of controls. b OR adjusted with sex and age. Table V. Association of COX-2 1195G.A genotype with CRC disease status Study group OR (95% CI) a according to COX-2 1195G.A genotype reported regarding the COX-2 765G.C polymorphism and risk of CRC and thus provide comparison with our findings. One study in Singapore Chinese showed that increased risk of colon cancer was associated with the 765C allele (27), which is in agreement with our results. The other study in American Caucasians, however, reported a reduced risk of colorectal adenoma associated with the 765C allele but the risk reduction was found only among non-users of NSAIDs (28). The inconsistent results among different ethnic populations might be associated with very different allele frequencies of COX-2 765C in Caucasians (17,28) and in our study population. Alternatively, it might indicate that the COX-2 765G.C polymorphism plays a role in carcinogenic processes under specific conditions or in specific populations. The 765G.C change seems to be functionally relevant. It has been reported that this polymorphism may eliminate an Sp1-binding site, resulting in reduced COX-2 expression (15,17). However, other study indicated that this polymorphism may create an E2F-binding site and the CC genotype is associated with a 10-fold higher prostaglandin production compared with the GG genotype (16). It would be interesting to examine the effect of COX-2 765G.C polymorphism on COX-2 expression under different ethnic backgrounds. To our best knowledge, thus far, there have been no published reports regarding the COX-2 1195G.A polymorphism and risk of CRC. We have recently shown that the 1195G.A change creates a transcriptional factor c-myb-binding site and the 1195A allele displays a higher promoter activity and higher COX-2 mrna expression compared with the 1195G allele (14). Our molecular epidemiological findings in the present study are consistent with biological function of the 1195G.A polymorphism. We also observed a positive association between the 12-LOX 261Gln/Gln genotype and moderately increased risk of CRC. This observation is biologically plausible because functional analysis showed that the 12-LOX 261Gln/Gln genotype carriers have significantly higher platelet 12-LOX activity towards arachidonic acid to form 12-HETE compared with the 12-LOX Arg/Arg genotype carriers GG (no. of subjects) GA (no. of subjects) AA (no. of subjects) GA þ AA (no. of subjects) Tumor stage Dukes B versus Dukes A 1.00 (reference) (65/30) 1.75 (1.00 3.15) (191/50) 1.60 (0.88 3.01) (115/33) 1.69 (1.00 2.89) (306/83) Dukes C versus Dukes A 1.00 (reference) (58/30) 1.82 (1.02 3.22) (174/50) 1.66 (0.88 3.16) (107/33) 1.74 (1.02 3.00) (281/83) Dukes D versus Dukes A 1.00 (reference) (20/30) 2.43 (1.15 4.97) (79/50) 2.66 (1.23 5.74) (58/33) 2.47 (1.25 4.89) (137/83) Tumor grade High versus low 1.00 (reference) (18/17) 0.85 (0.36 1.93) (57/64) 0.85 (0.33 1.96) (42/48) 0.82 (0.37 1.83) (99/112) High versus low þ intermediate 1.00 (reference) (18/150) 1.15 (0.63 2.15) (57/411) 1.44 (0.78 2.65) (42/250) 1.24 (0.70 2.24) (99/661) a OR adjusted with sex and age. 1200 (19). However, null association between this polymorphism and risk of CRC was also reported in Caucasians and African-Americans (25). The reason for this discrepancy is not clear, but might be associated with ethnic differences in allele and genotype frequencies. In this study with 1300 controls, we found that the frequencies of 12-LOX 261Gln allele and 261Gln/Gln genotype were 0.49 and 23.7%, whereas in Caucasians and African-Americans, they were 0.60 and 37.0% and 0.64 and 41.4%, respectively (25). In addition, statistical power should also be considered as an influencing factor for the discrepancy. Another interesting finding in the present study is that the COX-2 1195A allele was also associated with increased risk for developing higher Dukes stages of CRC at the time of diagnosis. These results might demonstrate an important role of the polymorphism as a relevant genetic factor for the progression of CRC. Cancer invasiveness and metastasis are the major cause of death in patients with the disease. Although it is not fully understood how primary CRC evolves to invasive and metastatic disease, many studies have shown that the over-expression of COX-2 is correlated with colon cancer cell invasion and advanced stages of CRC (29 33). It has been known that COX-2 and its downstream products prostaglandins play important roles in stimulating cell proliferation and angiogenesis and suppressing immune function, all of which may facilitate tumor cells to spread. Taken together, these findings indicate that COX-2 may play its role not only in colorectal tumorigenesis but also in the cancer progression. Because of these effects of COX-2 on CRC and gain of function of the 1195G.A polymorphism, it is biologically plausible to observe such an association between the COX-2 1195A allele and the risk for developing advanced Dukes stages of CRC. This information might be helpful in patient selection for more effective chemotherapy or radiotherapy. However, we did not find any effect of 12-LOX 261Arg.Gln polymorphism on CRC disease status, indicating that this polymorphism is unlikely to play a role in CRC progression.
Associations of COX-2 and 12-LOX polymorphisms with colorectal cancer Because both COX-2 and 12-LOX are involved in the arachidonic acid metabolism to form prostaglandins and HETEs that affect colorectal carcinogenesis, an additive or multiplicative interaction between COX-2 1195AA and 12-LOX 261Gln/Gln genotypes in intensifying risk of developing CRC had been presumed. However, the observed genotyping data fitted neither of the statistical models. In contrast, subjects carrying both variant genotypes with higher COX-2 expression and higher 12-LOX activity had an observed OR that was lower than the expected OR in terms of multiplicative scale. Because COX-2 and 12-LOX proteins compete for the same substrate arachidonic acid, it might make sense that the effects are less than would be expected if they were independent. The present study has some limitations. The risk of CRC related to COX-2 and 12-LOX polymorphisms might be influenced by use of NSAIDs and other environmental risk factors of the cancer such as alcohol intake. These factors might interact with COX-2 and 12-LOX genotypes or act as potential confounders in the analysis. Unfortunately, information on these factors in the present study is unavailable. It would be important in the future to investigate the interaction between the genotype and these factors, especially NSAID use because the effects of polymorphisms in COXs and some LOXs seem to differ depending on anti-inflammatory drug use (24,27,28,34 36). Another limitation is that we only analyzed promoter polymorphisms in the present study. It would be interesting and important to explore genetic variability in the entire COX-2 gene and risk of CRC. In addition, because our cases were recruited from a tertiary center and controls from a screening program, biases might occur with respect to extensiveness of disease and tumor growth characteristics and representative of the population from which the cases arose. In conclusion, our study demonstrates that the COX-2 1195G.A, COX-2 765G.C and 12-LOX 261Arg.Gln polymorphisms are genetic susceptibility factors for developing CRC in Han Chinese population. Our study also demonstrates that the COX-2 1195A allele is associated with increased risk for the development of advanced stages of CRC at the time of diagnosis. These results are consistent with the biological functions of the polymorphisms and support the hypothesis that aberrant polyunsaturated fatty acid metabolism may play a role in colorectal carcinogenesis and progression. Acknowledgements This study was supported by grants from National Natural Science Foundation (No. 30471957), National Basic Research Program (No. 2004CB518701 and No. 2002CB512902) and Program for New Century Excellent Talents in University (NCET-05-0171). References 1. Romano,M. et al. (2003) Cyclooxygenase-2 and 5-lipoxygenase converging functions on cell proliferation and tumor angiogenesis: implications for cancer therapy. FASEB J., 17, 1986 1995. 2. Jones,R. et al. (2003) Arachidonic acid and colorectal carcinogenesis. Mol. Cell. Biochem., 253, 141 149. 3. Williams,C.S. et al. (1999) The role of cyclooxygenases in inflammation, cancer, and development. Oncogene, 18, 7908 7916. 4. Gasparini,G. et al. (2003) Inhibitors of cyclooxygenase 2: a new class of anticancer agents? Lancet Oncol., 4, 605 615. 5. Eberhart,C.E. et al. (1994) Up-regulation of cyclooxygenase 2 gene expression in human colorectal adenomas and adenocarcinomas. Gastroenterology, 107, 1183 1188. 6. Reddy,B.S. et al. (1992) Inhibition of colon carcinogenesis by prostaglandin synthesis inhibitors and related compounds. Carcinogenesis, 13, 1019 1023. 7. Gupta,R.A. et al. (2001) Colorectal cancer prevention and treatment by inhibition of cyclooxygenase-2. Nat. Rev. Cancer, 1, 11 21. 8. Thun,M.J. et al. (1991) Aspirin use and reduced risk of fatal colon cancer. N. Engl. J. Med., 325, 1593 1596. 9. Brash,A.R. (1999) Lipoxygenases: occurrence, functions, catalysis, and acquisition of substrate. J. Biol. Chem., 274, 23679 23682. 10. Shureiqi,I. et al. (2001) Lipoxygenase modulation to reverse carcinogenesis. Cancer Res., 61, 6307 6312. 11.Yoshimoto,T. et al. (2002) Arachidonate 12-lipoxygenases. Prostaglandins Other Lipid Mediat., 68 69, 245 262. 12. Szekeres,C.K. et al. (2002) 12(S)-HETE, pleiotropic functions, multiple signaling pathways. Adv. Exp. Med. Biol., 507, 509 515. 13. Kamitani,H. et al. (1999) The possible involvement of 15-lipoxygenase/ leukocyte type 12-lipoxygenase in colorectal carcinogenesis. Adv. Exp. Med. Biol., 469, 593 598. 14. Zhang,X.M. et al. (2005) Identification of functional genetic variants in cyclooxygenase-2 and their association with risk of esophageal cancer. Gastroenterology, 129, 565 576. 15. Papafili,A. et al. (2002) Common promoter variant in cyclooxygenase-2 represses gene expression: evidence of role in acute-phase inflammatory response. Arterioscler. Thromb. Vasc. Biol., 22, 1631 1636. 16. Szczeklik,W. et al. (2004) Functional effects and gender association of COX-2 gene polymorphism G -765 C in bronchial asthma. J. Allergy Clin. Immunol., 114, 248 253. 17. Brosens,L.A. et al. (2005) Increased cyclooxygenase-2 expression in duodenal compared with colonic tissues in familial adenomatous polyposis and relationship to the 765G/C COX-2 polymorphism. Clin. Cancer Res., 11, 4090 4096. 18. Fridman,C. et al. (2003) Association of a new polymorphism in ALOX12 gene with bipolar disorder. Eur. Arch. Psychiatry Clin. Neurosci., 253, 40 43. 19. Guo,Y.L. et al. (2006) Platelet 12-lipoxygenase Arg261Gln polymorphism: functional characterization and association with risk of esophageal squamous cell carcinoma in combination with COX-2 polymorphisms. Pharmacogenet. Genomics, in press. 20. Jass,J.R. et al. (1989) Histological typing of intestinal tumours. In Jass,J.R. and Sobin,L.H. (eds) International Histological Classification of Tumours, 2nd edn. Springer-Verlag, New York, pp. 29 40. 21. Schaid,D.J. et al. (2002) Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am.J.Hum.Genet., 70, 425 434. 22. Lake,S.L. et al. (2003) Estimation and tests of haplotype-environment interaction when linkage phase is ambiguous. Hum. Hered., 55, 56 65. 23. Lin,H.J. et al. (2002) Prostaglandin H synthase 2 variant (Val511Ala) in African Americans may reduce the risk for colorectal neoplasia. Cancer Epidemiol. Biomarkers Prev., 11, 1305 1315. 24. Sansbury,L.B. et al. (2006) COX-2 polymorphism, use of nonsteroidal antiinflammatory drugs, and risk of colon cancer in African Americans (United States). Cancer Causes Control, 17, 257 266. 25. Goodman,J.E. et al. (2004) Arachidonate lipoxygenase (ALOX) and cyclooxygenase (COX) polymorphisms and colon cancer risk. Carcinogenesis, 25, 2467 2472. 26. Cox,D.G. et al. (2004) Polymorphisms in prostaglandin synthase 2/cyclooxygenase 2 (PTGS2/COX2) and risk of colorectal cancer. Br. J. Cancer, 91, 339 343. 27. Koh,W.P. et al. (2004) Interaction between cyclooxygenase-2 gene polymorphism and dietary n-6 polyunsaturated fatty acids on colon cancer risk: the Singapore Chinese Health Study. Br. J. Cancer, 90, 1760 1764. 28. Ulrich,C.M. et al. (2005) PTGS2 (COX-2) 765G.C promoter variant reduces risk of colorectal adenoma among nonusers of nonsteroidal antiinflammatory drugs. Cancer Epidemiol. Biomarkers Prev., 14, 616 619. 29. Tomozawa,S. et al. (2000) Cyclooxygenase-2 overexpression correlates with tumour recurrence, especially haematogenous metastasis, of colorectal cancer. Br. J. Cancer, 83, 324 328. 30. Zhang,H. et al. (2002) Overexpression of cyclooxygenase-2 correlates with advanced stages of colorectal cancer. Am. J. Gastroenterol., 97, 1037 1041. 31. Konno,H. et al. (2002) Cyclooxygenase-2 expression correlates with upar levels and is responsible for poor prognosis of colorectal cancer. Clin. Exp. Metastasis, 19, 527 534. 32. Yao,M. et al. (2004) Cyclooxygenase-2 selective inhibition with NS-398 suppresses proliferation and invasiveness and delays liver metastasis in colorectal cancer. Br. J. Cancer, 90, 712 719. 33. Tsujii,M. et al. (1997) Cyclooxygenase-2 expression in human colon cancer cells increase metastatic potential. Proc. Natl Acad. Sci. USA, 94, 3336 3340. 34. Ulrich,C.M. et al. (2004) Polymorphisms in PTGS1 (5 COX-1) and risk of colorectal polyps. Cancer Epidemiol. Biomarkers Prev., 13, 889 893. 35.Ulrich,C.M. et al. (2006) Non-steroidal anti-inflammatory drugs for cancer prevention: promise, perils and pharmacogenetics. Nat. Rev. Cancer, 6, 130 140. 36. Poole,E.M. et al. (2006) Prostacyclin synthase and arachidonate 5-lipoxygenase polymorphisms and risk of colorectal polyps. Cancer Epidemiol. Biomarkers Prev., 15, 502 508. Received September 17, 2006; revised November 27, 2006; accepted November 30, 2006 1201