C-reactive protein genotypes and haplotypes, polymorphisms in NSAID-metabolizing enzymes, and risk of colorectal polyps

Abstract

Introduction

C-reactive protein (CRP) is a non-specific marker of inflammation linked to cardiovascular disease and possibly colon cancer. Polymorphisms in CRP have been associated with differential CRP concentrations among healthy adults, with some evidence for functional effects on CRP expression.

Methods

A linkage-disequilibrium-based tagSNP-selection algorithm identified six tagSNPs for Europeans (−821A>G, −390C>T/A 90A>T 838G>C 2043G>A and 4363C>A), defining 6 haplotypes >1% frequency. In a case-control study of adenomatous (n=491) or hyperplastic (n=184) polyps vs. polyp-free controls (n=583) we investigated these SNPs in relation to colorectal polyp risk.

Results

Individuals with 838 GC or CC genotypes had a modestly, although not statistically significantly, increased risk of adenomas (OR=1.4 95% CI 0.9-2.1) and a nearly 2-fold increased risk of concurrent adenomas and hyperplastic polyps (OR=2.0 95% CI 1.1-3.6). Increased risk for concurrent adenomas and hyperplastic polyps was also observed for haplotype ACACAC. No other main associations were detected. Risk of adenomas associated with 2043G>A differed with NSAID use. Among NSAID non-users, there was a suggestion that the GA or AA genotypes were associated with decreased risk of adenomas; this was not seen among NSAID users (p-interaction = 0.03). We also observed interactions between UGT1A1 [TA](7) promoter repeat polymorphism and CRP tagSNPs −390C>T/A and 90A>T, in which only the homozygous variant CRP genotype was associated with increased adenoma risk among those with the UGT1A1 6rpt/6rpt genotype (p-interaction= 0.02 and 0.04 for −390C>T/A and 90A>T, respectively).

Conclusions

These results provide limited support for associations between genetic variation in CRP and colorectal polyp risk. The observed interactions should be evaluated further.

Introduction

C-reactive protein (CRP) is a non-specific acute-phase protein secreted by the liver in response to pro-inflammatory cytokines such as IL-6. Elevated CRP levels have been associated with increased risk of colorectal cancer in four prospective studies [1–4], but not in two others [5, 6]. Genetic polymorphisms in CRP have been associated with changes in CRP serum or plasma concentrations [7–15]. However, several of these polymorphisms are in intronic or untranslated regions of the CRP gene and associations with CRP concentrations may be attributable to a linked, but not yet identified, polymorphism.

Non-steroidal anti-inflammatory drugs (NSAIDs), including COX-2 specific NSAIDs (coxibs), have been consistently associated with a reduced risk of colorectal neoplasia (reviewed in [16]) and have been shown in clinical trials to prevent polyp recurrence [17–21]. NSAIDs have also been shown in some, but not all, studies to reduce CRP concentrations (reviewed in [22]). Thus, the reduction of risk in colorectal neoplasia afforded by NSAIDs may be partly through or marked by their effects on CRP concentrations (see Figure).

Figure.

CRP and NSAID use in inflammation.

NSAIDs are primarily metabolized by oxidation or by glucuronidation. The former is achieved primarily by cytochrome P540 2C9 (CYP2C9) [23] and the latter by the UDP-glucoronosyltransferases (UGTs), such as UGT1A1, UGT1A6, UGT1A9, UGT2B4, and UGT2B7 [24]. Polymorphisms in several of these enzymes have been shown to alter drug metabolism [25–29]. Additionally, polymorphisms in CYP2C9 and UGT1A6 have been associated with risk of colorectal polyps or cancer and shown to interact with NSAID use in some, but not all, studies [30–33]. Because functional polymorphisms in NSAID-metabolism genes may alter the ability of NSAIDs to reduce CRP concentrations, we hypothesized that polymorphisms in NSAID metabolism may be effect modifiers of the relationship between genetic variation in CRP and risk of colorectal neoplasia (see Figure).

We selected tagSNPs to capture common CRP haplotypes (≥ 1% haplotype frequency); these haplotypes were analyzed in a colonoscopy-based case-control study of colorectal polyps. We also investigated whether CRP genotype or haplotype associations differed by NSAID use or by polymorphisms in genes encoding NSAID-metabolizing enzymes (UGTs and CYP2C9).

Materials and Methods

Participant recruitment has been previously described [30, 34, 35]. Briefly, adenoma and hyperplastic polyp cases and polyp-free controls were recruited through a large multiclinic gastroenterological practice in the Twin cities area of Minnesota from April 1991-April 1994. Eligibility criteria have been described elsewhere [34]; participants were aged 30-74 years, English-speaking residents of the Twin Cities metropolitan area with no known genetic syndrome associated with increased risk of colon neoplasia and no individual history of cancer (except non-melanoma skin cancer), prior colorectal polyps, or inflammatory bowel disease. Information on use of aspirin or other NSAIDs, diet, physical activity, anthropometrics, demographics, and medical history was obtained via questionnaire. The participation rate for all colonoscoped patients was 68%.

TagSNP selection

The CRP coding region and 2KB 5’ and 3’ of the gene was resequenced by the University of Washington-Fred Hutchinson Cancer Research Center Variation Discovery Resource (UW-FHCRC VDR) [36]. TagSNPs were selected using the LD Select algorithm developed by Carlson and colleagues [37] at the UW-FHCRC VDR. However, we used more stringent criteria specifically, a minor allele frequency of 4% (i.e. any variant that occurred twice in the resequencing effort) and an r² value of 0.90. This resulted in the selection of six tagSNPs estimated by the Genome Variation Server (http://gvs.gs.washington.edu/GVS/index.jsp) to cover 85% of the variation in the CRP locus: −821A>G (rs2794521), −390C>T/A (rs3091244), 90A>T (rs1417938), 838G>C (L184L, rs1800947), 2043G>A (rs1205), and 4363C>A (rs3093075).

Genotyping

The selected CRP tagSNPs were detected by allelic discrimination using the 5’ nuclease assay on a 7900HT sequence detection system (Applied Biosystems, Foster City, CA). The 5’nuclease genotyping assays were validated by genotyping 92 individuals by both 5’ nuclease assay and RFLP or sequencing. There were no discrepancies between the two assays. The 20μl genotyping reactions contained 1x Taqman Core Reagents (Applied Biosystems), except for −821A>G for which the universal master mix was used, 0.5 units AmpliTaq DNA polymerase, 0.2 units AmpErase UNG, primers, probes, and 4ng genomic DNA. Primers, probes, Mg²⁺ concentrations, and cycling conditions are listed in Table 1. The triallelic polymorphism at −390 (−390C>T/A ) was run as a real-time assay and all the other polymorphisms as end-point assays. Positive controls for all the genotypes as well as four negative controls were included on each plate. For quality control purposes, genotyping for 94 randomly selected samples was repeated. There were no discrepancies.

Table 1.

PCR Conditions

Polymorphism	PCR	Primers/Probes	[Mg2+]	[Primers, Probes]	Amplicon	Cycling
−821A>G	FP	5'GGCCGTCATTTAGTGCCAA3'		200nM		50°C, 2min, 95°C, 10min, 40x 94°C, 30sec; 58°C, 45sec; 72°C, 1min 72°C, 5min
	RP	5'GTGCTGCACCCATTAACTCATC3'		200nM
	A-allele	5’6FAM-CACCGCATGTTCT-3’MGB	2.5mM	100nM	250bp
	G-allele	5'VIC-CACCGCGTGTTCT-3’MGB		100nM
−390C>T/A	FP	5'TCAGATTTCCTTTGTCAAACTCTATGA3'		200nM		50°C, 2min, 95°C, 10min, 40x 95°C, 15sec; 60°C, 90sec
	RP	5'TCCACTTTGGCTATCTATCCTGC3'		200nM
	C-allele	5’VIC-AACATATTAAACGAGTGGCCAT-3’MGB	4mM	100nM	138bp
	T-allele	5’TET-CATATTAAACAAGTGGCCATC-3’MGB		100nM
	A-allele	5’6FAM-TAACATATTAAACTAGTGGCCATC-3’MGB		100nM
90A>T	FP	5'TGGCCAGACAGGTAAGGGC3'		200nM		50°C ,2min, 95°C, 10min, 40x 95°C, 15sec; 59°C, 90sec
	RP	5'ACCATGAAGGATGCTCCACTG3'		200nM
	A-allele	5’VIC-TCAGATCAAATCTCTCCCAT-3’MGB	4mM	100nM	141bp
	T-allele	5’6FAM-TCAGATCAAAACTCTCCCATA-3’MGB		100nM
838G>C	FP	5'TGGGAGACATTGGAAATGTGAAC3'		200nM		50°C, 2min, 95°C, 10min, 40x 95°C, 15sec; 59°C, 1 min
	RP	5'CCGCCAAGATAGATGGTGTTAAT3'		200nM
	G-allele	5’VIC-TTGTGCTGTCACCAGA-3’MGB	3mM	100nM	76bp
	C-allele	5’6FAM-TTTGTGCTCTCACCAGA-3’MGB		100nM
2043G>A	FP	5'GCCATCTTGTTTGCCACATG3'		200nM		50°C, 2min, 95°C, 10min, 40X 95°C, 15sec; 60°C, 1 min
	RP	5'CCCTTGGCTCCTCCACTTC3'		200nM
	G-allele	5'VIC-TGTCCTCACAGTCTC-3’MGB	5mM	100nM	70bp
	A-allele	5'6-FAM-TGTCCTCATAGTCTCT-3’MGB		150nM
4363C>A	FP	5'AACCTAAAATCTCCCTGTGTCAGAA3'		200nM		50°C, 2min, 95°C, 10min, 40x 95°C, 15sec; 61°C, 2min 30sec
	RP	5'CTACTTACTTTGTCAGCTGGGACTCC3'		200nM
	C-allele	5’VIC-TTTCCATCAGGTCCCA-3’MGB	4mM	100nM	263bp
	A-allele	5'6-FAM-TCCATCATGTCCCAGC-3’MGB		100nM

Genotyping of CYP2C9, UGT1A1, and UGT1A6 has been described previously [27, 30]. Briefly, polymorphisms in CYP2C9 and UGT1A6 were genotyped by restriction fragment length polymorphism and oligonucleotide ligation assay. The UGT1A1 polymorphism was genotyped by PCR.

Haplotypes

Haplotypes were inferred using proc haplotype in SAS Genetics v.9 (SAS Institute, Cary, NC). All haplotypes predicted to occur with more than 1% frequency were analyzed separately and the haplotypes with lower frequency were grouped together for analyses. The most common haplotype among the controls was used as the referent group.

Statistical Methods

Three cases groups were defined: adenomas (n=477, hyperplastic polyps (n=177, and those with concurrent adenomas and hyperplastic polyps (n=112). Cases with concurrent adenomas and hyperplastic polyps were included in the adenoma and hyperplastic analyses.

Unconditional logistic regression was used to estimate odds ratios (ORs) and corresponding 95% confidence intervals (CI) for the associations between CRP genotypes and haplotypes and polyp risk. For SNPs with a minor allele frequency <10%, we grouped the homozygous variant genotypes with the heterozygous genotypes. A logistic regression model using GEE was used to analyze haplotype effects, using the haplotype probabilities as weights and clustering the haplotypes for each individual [38]. Global tests of haplotype associations were calculated using score tests in the genmod procedure. Covariates considered in our models included age, sex, body mass index, dietary intakes of fiber, alcohol, and energy, postmenopausal hormone use, and smoking. Effect modification by NSAID use was evaluated by the inclusion of multiplicative interaction terms in logistic regression models. Because use of NSAIDs may be associated with other known risk factors for colorectal neoplasia, we adjusted our NSAID-interaction analyses for the variables listed above. All statistical analyses were carried out using SAS v.9.

Results

A total of 1719 subjects were recruited into this study, of which 1217 had DNA available for genotyping. Characteristics of the study population have been described previously [30, 34, 35]. Briefly, the study population was mostly Caucasian (97.2%) and tagSNPs were selected within that group; adenoma cases tended to be older than hyperplastic polyp cases and controls. Both sets of cases were more likely to be male than controls. Genotype frequencies for all polymorphisms were in Hardy-Weinberg Equilibrium among the controls.

Risk of colon polyps associated with CRP genotypes and haplotypes are presented in Table 2 and in Supplemental Table S1 online. There were no differences in risk of adenomas or hyperplastic polyps for any of the genotypes, except possibly 838G>C, for which there was a non-significant increase in risk of adenoma (OR 1.4, 95%CI 0.9-2.1) and a marginally significant increased risk of concurrent adenomatous and hyperplastic polyps associated with the C allele (OR 2.0, 95%CI 1.1-3.6). This was no longer significant when Bonferroni correction for multiple testing was applied. Similarly, global tests of the haplotype associations were statistically non-significant (see Tables 2 and S1). There were no associations with adenoma risk observed for any of the individual imputed haplotypes; however the ACACAC haplotype and the grouping of all rare haplotypes were both associated with increased risk of concurrent adenomas and hyperplastic polyps. Multivariate adjustment for BMI, fiber intake, total energy intake, alcohol, hormone use (women), and smoking did not alter odds ratio estimates; thus results adjusted for age and sex are presented. Tests for heterogeneity of odds ratios, using the contrast statement in multinomial regression in SAS (the logistic procedure with the glogit link specified) in adenoma models vs. concurrent adenomas and hyperplastic polyps were statistically non-significant (p=0.86).

Table 2.

Risk of colorectal polyps associated with CRP genotypes¹

	Controls (N=562)	Adenomas (N=477)		Concurrent adenomas and hyperplastic polyps (N=112)
	N	N	OR (95% CI)	N	OR (95% CI)
CRPGenotype
−821A>G rs2794521
AA	294	245	1.0 (ref.)	61	1.0 (ref.)
AG	217	180	1.0 (0.7-1.3)	41	0.8 (0.5-1.3)
GG	51	53	1.4 (0.9-2.1)	11	1.2 (0.5-2.5)
−390C>T/A rs3091244 (all genotypes)
CC	221	196	1.0 (ref.)	43	1.0 (ref.)
CT	215	181	1.0 (0.7-1.3)	43	1.1 (0.6-1.8)
CA	56	37	0.9 (0.6-1.5)	12	1.5 (0.7-3.2)
TT	42	42	1.1 (0.7-1.8)	10	1.1 (0.5-2.4)
AA	4	1	0.2 (0.1-1.4)	0	0 (--)
TA	25	20	0.9 (0.4-1.6)	4	0.8 (0.3-2.7)
−390C>T/A rs3091244 (grouped by putative phenotype as in [15])
CC	221	196	1.0 (ref.)	43	1.0 (ref.)
CT or CA	271	218	1.0 (0.7-1.3)	55	1.2 (0.7-1.7)
TT, AA, or TA	71	63	1.0 (0.6-1.4)	14	0.9 (0.4-1.8)
90A>T rs1417938
AA	281	236	1.0 (ref.)	57	1.0 (ref.)
AT	239	200	1.0 (0.8-1.3)	46	1.0 (0.6-1.5)
TT	43	42	1.1 (0.6-1.8)	10	1.0 (0.4-2.1)
838G>C rs1800947 (L184L)
GG	504	414	1.00(ref.)	94	1.0 (ref.)
GC or CC	59	64	1.4 (0.9-2.1)	19	2.0 (1.1-3.6)
2043G>A rs1205
GG	241	218	1.0 (ref.)	50	1.0 (ref.)
GA	262	210	0.9 (0.7-1.2)	51	1.0 (0.6-1.7)
AA	60	50	0.9 (0.6-1.4)	12	1.1 (0.5-2.2)
4363C>A rs3093075
CC	477	419	1.0 (ref.)	96	1.0 (ref.)
CA or AA	86	57	0.9 (0.6-1.3)	16	1.1 (0.6-2.0)
CRPhaplotype2(−821A>G; −390C>T/A; 90A>T; 838G>C; 2043G>A; 4363C>A)
GCAGGC	28.6	29.3	1.0 (ref.)	26.9	1.0 (ref.)
ATTGGC	28.4	29.7	1.0 (0.9-1.2)	30.0	1.1 (0.8-1.4)
ACAGGC	1.3	1.9	1.2 (0.8-1. 8)	2.6	1.5 (0.8-2.8)
ACAGAC	28.1	25.1	0.9 (0.8-1.1)	24.3	1.0 (0.7-1.3)
ACACAC	5.2	7.1	1.2 (0.9-1.6)	7.8	1.6 (1.0-2.5)
AAAGGA	8.1	6.3	0.9 (0.7-1.2)	7.0	1.1 (0.7-1.78)
all rare haplotypes	0.4	0.8	1.5 (0.7-3.1)	1.3	2.0 (1.0-3.8)
			global p=0.63		global p=0.49

¹Age and sex adjusted.

²Percents rather than total N are reported for haplotypes since they are inferred rather than determined.

We detected a significant interaction between regular aspirin or other NSAID use and the 2043G>A polymorphism (Table 3). Among those with the common allele for 2043G>A (i.e. GG), NSAID use was associated with a decreased risk of adenoma (OR 0.4, 95% CI 0.3-0.7), whereas among those with at least one A allele, NSAID use was not associated with a further decrease in adenoma risk (AG: OR 0.6, 95% CI 0.4-0.9; AA: OR 0.5, 95% CI 0.4-1.2; p-interaction = 0.03). However, when Bonferroni correction is applied for multiple testing, the p-value for this interaction is 0.18. No other NSAID interactions were observed for CRP tagSNPs (see Table S2 online). For CRP haplotypes, no interactions with NSAID use were observed (p-interaction=0.63 for adenomas). Because BMI is an important predictor of CRP levels, we also investigated potential interactions between BMI and CRP genotypes. No statistically significant interactions were observed (data not shown).

Table 3.

Association between CRP 2043G>A and risk of adenoma, stratified by NSAID use^*

Common Referent Group								Separate Referent Groups
	Aspirin or other NSAID use							Aspirin or other NSAID use
	No			Yes				No			Yes
	Controls	Cases		Controls	Cases			Controls	Cases		Controls	Cases
	(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)		(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)
2043G>A rs1205
GG	124	145	1.0 (ref.)	117	73	0.4 (0.3-0.7)		124	145	1.0 (ref.)	117	73	1.0 (ref.)
GA	154	125	0.6 (0.4-0.9)	108	85	0.6 (0.4-0.9)		154	125	0.6 (0.4-0.9)	108	85	1.4 (0.9-2.3)
AA	34	33	0.7 (0.4-1.4)	26	17	0.5 (0.2-1.1)		34	33	0.7 (0.4-1.4-)	26	17	1.2 (0.5-2.6)
													p-interaction = 0.03

^*Adjusted for age, sex, BMI, intakes of total calories, alcohol, and fiber, smoking (pack-years) and hormone use (females).

Statistically significant interactions were observed between two CRP polymorphisms (−390C>T/A and 90A>T) and the UGT1A1 promoter TA repeat polymorphism. Among those who were homozygous variant for either CRP genotype, the most frequent UGT1A1 genotype (6rpt/6rpt) was associated with increased risk of adenoma, whereas having an increased number of UGT1A1 repeats was associated with decreased risk (Table 4). No other interactions between CRP and UGT or CYP2C9 polymorphisms were observed (see Tables S3–S5 online).

Table 4.

CRP genotype, UGT1A1 genotype, and risk of colorectal adenoma¹

	Common Referent Group							Separate Referent Groups
	UGT1A1 genotype							UGT1A1 genotype
	6/6 (wildtype)			6/7, 6/8, 7/7, 7/8				6/6 (wildtype)			6/7, 6/8, 7/7, 7/8
	Controls	Cases		Controls	Cases			Controls	Cases		Controls	Cases
CRP Genotype	(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)		(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)
−390C>T/A rs3091244 (grouped by putative phenotype)
CC	101	92	1.0 (ref.)	117	103	1.1 (0.7-1.7)		101	92	1.0 (ref.)	117	103	1.0 (ref.)
CT or CA	114	91	1.0 (0.7-1.6)	154	126	1.1 (0.7-1.6)		114	91	1.0 (0.7-1.6)	154	126	1.0 (0.7-1.4)
TT, AA, or TA	25	32	2.1 (1.0-4.2)	45	30	0.7 (0.4-1.2)		25	32	2.1 (1.0-4.2)	45	30	0.6 (0.3-1.1)
					p-interaction = 0.02					p-trend=0.08			p-trend=0.19
90A>T rs1417938
AA	120	105	1.0 (ref.)	156	130	1.1 (0.7-1.6)		120	105	1.0 (ref.)	156	130	1.0 (ref.)
AT	104	86	1.1 (0.7-1.7)	133	112	1.0 (0.7-1.6)		104	86	1.1 (0.7-1.7)	133	112	1.0 (0.7-1.4)
TT	16	25	2.6 (1.2-5.6)	27	17	0.7 (0.3-1.4)		16	25	2.6 (1.2-5.6)	27	17	0.6 (0.3-1.2)
					p-interaction = 0.04					p-trend=0.07			p-trend=0.31

¹Adjusted for age, sex, BMI, intakes of total calories, alcohol, and fiber, current NSAID use, smoking (pack-years) and hormone use (females).

Because CYP2C9 and UGT1A6 are the major biotransformation enzymes involved in the metabolism of NSAIDs and polymorphisms in these genes have been previously found to alter associations of NSAID use with colorectal neoplasia [30–33], we investigated whether non-synonymous SNPs in CYP2C9 (*2 and *3 alleles) and UGT1A6 (T181A + R184S or R184S alone) combined with NSAID use interacted with CRP SNPs. For the UGT1A6 polymorphism, the variant alleles are associated with slower drug metabolism [39], thus regular NSAID users with one or more variant alleles are likely to have the highest NSAID concentrations and, perhaps, the lowest risk of colorectal polyps [30]. Based on this hypothesis, we defined a three-level variable based on putative NSAID exposure: those who were wildtype and who didn’t use NSAIDs (=high risk) were in one group, those with any variant allele and who were regular NSAID users (=low risk) in a second group and a third group contained all other combinations (=intermediate). A statistically significant interaction was observed for UGT1A6/NSAID use and CRP −390C>T/A. Among those who were wildtype for UGT1A6 and did not use NSAIDs, there was no association with the CRP −390C>T/A polymorphism. However, among those with at least one variant UGT1A6 allele who regularly used NSAIDs, the lowest risk of adenoma was among those with the CRP homozygous variant alleles (Table 5; p-interaction = 0.03, Bonferroni correction p=0.18). Similar results were observed for hyperplastic polyps, although the interaction was not statistically significant. We followed a similar procedure for CYP2C9; however no interactions were observed (see Tables S6 and S7 online).

Table 5a.

CRP genotype, UGT1A6 genotype and NSAID use and risk of colorectal adenoma (common referent group)¹

	UGT1A6 genotype/regular NSAID use
	wt and no NSAIDs			All others			181A + 184S or 184S and NSAID use
	Controls	Cases		Controls	Cases		Controls	Cases
	(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)
CRPgenotype
−390C>T/A rs3091244 (grouped by putative phenotype)
CC	61	56	1.0 (ref.)	111	98	1.0 (0.6-1.7)	49	43	1.0 (0.5-1.8)
CT or CA	68	55	1.0 (0.6-1.8)	132	118	1.1 (0.7-1.8)	70	46	0.8 (0.5-1.5)
TT, AA, or TA	12	17	1.9 (0.7-5.0)	32	37	1.5 (0.7-2.9)	27	9	0.3 (0.1-0.7)
									p-interaction = 0.03

¹Adjusted for age, sex, BMI, intakes of total calories, alcohol, and fiber, smoking (pack-years) and hormone use (females).

Discussion

There is growing interest in the role of CRP in colorectal neoplasia. Studies of plasma or serum CRP measures and colorectal neoplasia risk generally suggest that increasing CRP concentrations are associated with increased risk of colorectal cancer [1–4], although these results are not entirely consistent [5, 6]. Our findings suggest that genetic variability in CRP is unlikely to play a role in colorectal neoplasia risk, but that polymorphisms in CRP may be relevant to the pharmacogenetics of NSAIDs. Out of six tagSNPs tested for an association with risk of colorectal polyps, only one, 838G>C, was associated with increased risk of concurrent adenomatous and hyperplastic polyps. This SNP is a synonymous polymorphism in exon 2 of the CRP gene (L184L) and the G allele has been previously, although not consistently, associated with higher plasma CRP levels [8, 12, 13, 40, 41].

To our knowledge only one previous study has examined polymorphisms in CRP and risk of colorectal neoplasia [4]. In that study, no association was observed between three tagSNPs (rs1130864, rs1205, and rs3093068) and colorectal cancer risk. One of these SNPs, rs1205, was included in our study; we also found no association. The other two SNPs in that study are in LD (r² ≥ 90%) with two SNPs that we included in our study (rs1130864 is in LD with 90A>T and rs3093068 is in LD with 4363C>A); we also found no associations with these SNPs. No previous studies have examined potential interactions between variation in CRP and regular NSAID use. However, given the importance of aspirin and other NSAIDs for chemoprevention of cardiovascular disease and colorectal neoplasia, and the increasing use of CRP as a biomarker for inflammation, we suggest that potential interactions between NSAID use and CRP genetic variation should be considered in future etiologic studies of the associations between CRP and colorectal cancer risk.

In this exploratory study, we observed several interactions with NSAID use, NSAID-metabolizing enzymes, or the combination, suggesting that research into the role of CRP in cancer risk should take NSAID use into account. NSAID use has been reported to reduce CRP levels (reviewed in [22]). For the NSAID interaction, the OR among the homozygous variant NSAID users was higher than would be expected in NSAID use and CRP genotype were independent risk factors (expected: 0.7*0.4=0.28 observed: 0.5, p-0.05). This observed interaction suggests that NSAID use may not be beneficial to all, but rather may be more effective among people with higher underlying inflammation. We have previously reported on NSAID-gene interactions in genes relevant to prostaglandin synthesis [42–44] and functional polymorphisms in NSAID-metabolizing enzymes UGT1A6 and CYP2C9 have also been reported to alter the NSAID-colorectal neoplasia association [30–33]. Although our findings require confirmation, the current results add to the evidence that pharmacogenetic studies are necessary to truly understand which patients are most likely to benefit from NSAID chemoprevention and which are most likely to experience side effects.

Because we chose tagSNPs rather than candidate functional SNPs for this study, we had no prior hypotheses as to the association between CRP polymorphisms and colorectal polyp risk. However, several of these tagSNPs have been previously associated with plasma or serum CRP levels, indicating that at least one of them has functional effects. Most consistently, the T and A alleles of −390C>T/A have been previously associated with increased plasma CRP concentrations [13–15, 45, 46]. Our results do not support a role for this polymorphism in polyp risk, but we did observe an interaction between −390C>T/A and the combination of UGT1A6 genotype and regular NSAID use. In this interaction, those with two variant -390 alleles (i.e. AA, TT or AT) had a decreased risk in combination with having UGT1A6 variant alleles, which are associated with slower drug metabolism, and NSAID use. If the CRP alleles are associated with increased CRP production, then this is the group in which NSAID use, particularly among slow metabolizers, would be expected to be the most effective. This and other reported interactions require confirmation in other studies.

This was an initial exploratory study of the association between CRP SNPs and colorectal neoplasia. We performed 48 tests for interaction; at α=0.05, by chance, we would expect at least 2 to be significant. We recognize that our sample size was relatively small and that our findings may be false positives. However, because this study was meant to inform further investigations of CRP variability and neoplasia risk, we decided to report all significant and near-significant findings.

In summary, we found limited evidence for an association between genetic variability in CRP and colorectal polyp risk, yet these polymorphisms may interact with NSAID use or NSAID-metabolizing enzymes. Further research is required to confirm these findings in larger studies of colorectal neoplasia. A comprehensive investigation of the genetics of biologic pathways relevant to inflammation, including prostaglandin synthesis and pro- and anti-inflammatory cytokines will further our understanding of the role of inherited susceptibility in colorectal cancer risk. Clearly such studies require information on use of NSAIDs.

Table 5b.

CRP genotype, UGT1A6 genotype and NSAID use and risk of colorectal adenoma (separately by strata of UGT1A6 genotype and NSAID use)¹

	UGT1A6 genotype/regular NSAID use
	wt and no NSAIDs			All others			181A + 184S or 184S and NSAID use
	Controls	Cases		Controls	Cases		Controls	Cases
	(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)	(N)	(N)	OR (95% CI)
CRPgenotype
−390C>T/A rs3091244 (grouped by putative phenotype)
CC	61	56	1.0 (ref.)	111	98	1.0 (ref.)	49	43	1.0 (ref.)
CT or CA	68	55	1.0 (0.6-1.8)	132	118	1.0 (0.7-1.6))	70	46	0.9 (0.5-1.7)
TT, AA, or TA	12	17	1.9 (0.7-5.0)	32	37	1.4 (0.8-2.7)	27	9	0.3 (0.1-0.7)
			p-trend=0.28			p-trend=0.32			p-trend=0.02

¹Adjusted for age, sex, BMI, intakes of total calories, alcohol, and fiber, smoking (pack- years) and hormone use (females).