Abstract
Objective
The gene coding for C-reactive protein (CRP) is located on chromosome 1q23.2, which falls within a linkage region thought to harbor a systemic lupus erythematosus (SLE) susceptibility gene. Recently, two SNPs in the CRP gene (+838, +2043) have been shown to be associated with CRP levels and/or SLE risk in a British family-based cohort. The current study was done to confirm the reported association in an independent population-based case-control cohort, and also to investigate the impact of three additional CRP tagSNPs (-861, -390, +90) on SLE risk and serum CRP levels.
Methods
DNA from 337 white women who met the ACR criteria for definite (n = 324) or probable (n = 13) SLE and 448 white female healthy controls was genotyped for five CRP tagSNPs (-861, -390, +90, +838, +2043). Genotyping was performed using PCR-RFLP, pyrosequencing or TaqMan assays. Serum CRP levels were measured using ELISA. Association studies were performed using the χ2 distribution, Z-test, Fisher's exact test and ANOVA. Haplotype analysis was performed using EH software and haplo.stats package in R 2.1.2.
Results
While none of the SNPs were found to be associated with SLE risk individually, there was an association with the five-SNP haplotypes (p<0.000001). Three SNPs (-861, -390, +90) were found to significantly influence serum CRP level in SLE cases, both independently and as haplotypes.
Conclusion
Our data suggests that unique haplotype combinations in the CRP gene may modify the risk of developing SLE and influence circulating CRP levels.
The pathogenesis of systemic lupus erythematosus (SLE) is complex and multifactorial, involving interactions among multiple genes, hormones and several environmental factors. Even though the etiopathogenesis of SLE remains elusive, it is believed that impaired handling of antigen–antibody complexes and subsequent tissue deposition leading to release of inflammatory mediators and an array of inflammatory cells can induce a broad spectrum of clinical manifestation[1]. Among a range of factors that are thought to be contributing to the pathophysiology of SLE, chronic inflammation is thought to play a pivotal role in the pathogenesis of SLE.
Family and twin studies suggest that genetic factors play a significant role in the predisposition to SLE[2,3]. The estimated heritability of SLE in Caucasian is 66%[4]. Recent genome wide linkage analyses in multiplex SLE families have provided many chromosomal regions for exploration of disease-predisposing genes, including a region on the q-arm of chromosome 1[5]. The gene coding for C-reactive protein (CRP) is located at 1q23, which falls within the 1q23-43 region thought to harbor a susceptibility gene for SLE in multiple independent genome scans of both mice and humans[6-10]. The unique position of the CRP gene makes it a logical positional candidate gene to investigate as a susceptibility locus for SLE.
CRP is also a functional candidate gene based on the physiological activity of its products. CRP is an important liver-derived acute-phase protein that can increase up to 1000-fold in serum as a response to diverse stimuli such as infection or injury[11]. CRP has been shown to bind chromatin[12], histones[13] and apoptotic cells[14]. These unique characteristics of CRP are thought to contribute to its ability to modify the autoimmune disease phenotype by promoting the removal of necrotic and apoptotic cells and recruiting complement and FcyR-mediated effector pathways[15]. In the host, the increased clearance of apoptotic cells and their derived nuclear contents by phagocytic cells via CRP opsonization may prevent the development of potential nuclear antigen-specific autoimmune responses[14,16]. Recent in vivo studies have shown that lupus-prone BW mice carrying the CRP transgene had reduced proteinuria, lived longer than non-transgenic BW, and had delayed accumulation of IgM and IgG in their renal glomeruli[17]. Injecting CRP to another lupus strain mouse, NZB/NZW, also delayed the onset of high-grade proteinuria and prolonged survival[18]. CRP's autoimmunity prevention ability may come from its ability to prevent activation of autoreactive B cells by promoting clearance of antoantigens to non-antigen presenting sites[15].
Several studies have shown that CRP levels in SLE patents are abnormally elevated both in the absence and presence of infection[19-23]. The value of using CRP to monitor SLE disease activity has remained controversial given the inconsistent correlation between circulating CRP and disease activity from numerous studies[24-28]. The abnormal elevation pattern of CRP in SLE patients provided the first clinical clue that variation in the CRP may contribute to the pathogenesis of SLE. With CRP's unquestionable tie to inflammation, association with atherogenesis, its unique ability to modify the disease phenotypes of SLE, and its status as a positional candidate gene, CRP serves as a promising susceptibility gene for SLE.
Russell et al.[29] found basal levels of CRP to be influenced independently by 2 CRP polymorphisms (+838 & +2043), and the latter was also associated with SLE and antinuclear autoantibody production. They hypothesized that defective disposal of potentially immunogenic material, indicated by low basal CRP levels, may be a contributory factor in lupus pathogenesis. In the present study, we examined five tagSNPs, including +838 and +2043 both individually and as haplotypes to investigate the associations of CRP with SLE risk and serum CRP levels in SLE patients. We hypothesized that the variation in the CRP gene may contribute to the genetic susceptibility of SLE and may have impact on CRP levels in patients with SLE.
SUBJECTS AND METHODS
Subjects
A total of 337 white female SLE cases and 448 healthy female controls were included in this study. All cases were 18 years of age or older (mean age 43 ± 11 years, 40.3% post-menopausal, disease duration 10.13 ± 7.13) and were recruited from the Pittsburgh Lupus Registry. All subjects met the 1982 and 1997 American College of Rheumatology (ACR) criteria for definite (n = 324) or probable (n = 13) SLE[30,31] at the time of recruitment.
Participating subjects in this study have been seen either at the University of Pittsburgh Medical Center or by practicing rheumatologists in the Pittsburgh metropolitan area. The diagnosis of SLE was confirmed by a rheumatologist at the University of Pittsburgh (SM) prior to entry into the study. Since these patients are not exclusively from a tertiary referral center, they represent a spectrum of SLE that may be more reflective of a population-based sample.
Controls were race, sex and geographically matched and obtained from the Central Bank of Pittsburgh, and had no apparent history of SLE (mean age 45 ± 13 years, 100% Caucasian, 100% female). This study was approved by the University of Pittsburgh Institutional Review Board, and all subjects provided written informed consent.
SLE Clinical and Laboratory Characteristics
A subset of SLE cases (n = 237, mean age 44.26 ± 10.9, 40% post-menopausal) participating in a Cardiovascular Disease Study in SLE had high-sensitivity CRP data for the current genetic association study[32]. None of the 237 SLE cases with CRP measurement had any evidence of infection at the time of the study CRP level (logCRP range: -1.6 to 4.4).
CRP was measured using high sensitivity enzyme-linked immunoabsorbent assay and details of the assay are previously described[33,34]. SLE disease activity and cumulative damage were measured by the same physician (SM) in all patients, using the Systemic Lupus Activity Measure (SLAM)[35] and the Systemic Lupus International Collaborating Clinics (SLICC) damage index[36], respectively. Renal disease among SLE patients was defined using the ACR criteria, which requires (a) renal biopsy showing lupus nephritis, or (b) persistent proteinuria greater than 0.5 grams per day or greater than 3+ if quantification is not performed, or (c) evidence of cellular casts in the urine. Central nervous involvement (CNS) and arthritis among SLE patients were defined by the ACR criteria including history of seizure or psychosis due to SLE for the former. Two hundred and ninety five patients (87.2%) had arthritis, 88 (26.1%) patients had a diagnosis of SLE-renal disease and 30 (8.9%) patients had CNS involvement. Patients' mean SLICC score at the time of recruitment was 1.42 (SD: 1.76) and mean SLAM score was 6.29 (SD: 3.73).Additional measurements included anti-double stranded DNA, antiphospholipid antibodies, serum C3, and C4.
TagSNP Selection and Genotyping
Five informative tag SNPs were selected from a total of 31 known SNPs in the SeattleSNPs Program for Genomic Applications web site (http://pga.gs.washington.edu/education.html). SNPs -861 and -390 are located in the promoter region, SNP +90 is located in intron/exon boundary, +838 is a synonymous SNP present within exon 2, and +2043 maps in the 3' untranslated region. We have designated our SNPs based on their position relative to the ATG codon of the CRP translation site in the FASTA database. For clarification, reference numbers from the NCBI Entrez SNP database are provided for each of our six SNPs: -861 is rs3093059, -390 is rs3091244, +90 is rs1417938, +838 is rs1800947 and +2043 is rs1205.
Genotyping for +838 and +2043 was obtained using polymerase chain reaction restriction fragment length polymorphism (PCR-RFLP). SNPs -861 was genotyped using pyrosequencing assays. SNPs -390 and +90 were genotyped using TaqMan assays. TaqMan assays required a higher concentration of DNA that was not available in all subjects at the time of the experiment. Therefore genotyping was performed on 275 cases and 375 controls for the -390 and +90 SNPs. Our genotyping success rates on all SNPs ranged from 89% to 99% in the total sample. Ten percent of samples were re-genotyped for a second time and the concordance rate were > 99%.
Statistical Analyses
Allele frequencies were calculated by the allele counting method. Goodness of fit to Hardy–Weinberg expected proportions was examined by χ2 test. The pairwise linkage disequilibrium (LD) between markers was estimated using the D' method[37]. The differences in genotype frequencies between cases and controls were tested by Fisher's Exact test. Common haplotype frequency was estimated using the expectation-maximization algorithm in the EH software program[38] in both cases and controls. After detecting a significant association between CRP genetic variation and SLE risk, we performed follow-up tests to assess possible associations between CRP genetic variation and SLE clinical characteristics using either analysis of variance (ANOVA) for quantitative clinical characteristics of SLE (C3, C4, SLAM and SLICC) or logistic regression analysis for categorical variables (renal disease, joint inflammation, CNS involvement and antiphospholipid antibodies). Covariates adjusted for in the models included age, BMI, and smoking.
To assess the association between CRP genetic variation and serum CRP in SLE subjects, CRP values were log-transformed to reduce non-normality. The mean log-transformed CRP (logCRP) levels between different genotype groups were compared using ANOVA and adjusted for the effects of age, BMI, and smoking. We also conducted haplotype association tests with logCRP were conducted using the haplo.stats package for R[39], with age, BMI, and smoking included as covariates. Haplo.stats tests association by means of a generalized linear-regression framework that incorporates haplotype phase uncertainty by inferring a probability matrix of haplotype likelihoods for each individual (derived by use of the EH haplotype-inference algorithm) rather than by assignment of a most likely haplotype. All computations were performed using R version 2.1.2.
RESULTS
Association of CRP SNPs with SLE Risk
Of the total 785 subjects (337 cases and 448 controls) genotyped for five CRP SNPs, we repeated genotyping on 10% of the subjects for each SNP a second time and had higher than 99% concordance rate in all SNPs. No statistically significant deviations from Hardy–Weinberg equilibrium were found in any of the SNPs. Table 1 presents the genotype and allele frequencies in our cases and controls for the five CRP SNPs examined. Genotype and allele frequencies were not significantly different between cases and controls (using the p-value of 0.05) in any of the individual five SNPs examined.
Table 1.
Genotype and allele frequencies of CRP SNPs
| SNP | Genotype | SLE cases n (%) | Controls n (%) | p-value | Allele | SLE cases n (%) | Controls n (%) | p-value |
|---|---|---|---|---|---|---|---|---|
| −861 | TT | 287 (85.93) | 388(86.8) | 0.813 | T | 621 (93) | 834(93.3) | 0.802 |
| (rs3093059) | TC | 47(14.07) | 58(12.98) | C | 47(7) | 60 (6.7) | ||
| CC | 0(0) | 1 (0.22) | ||||||
| −390 | CC | 92 (38) | 135 (41) | 0.849 | C | 289 (59) | 416(61) | 0.565 |
| (rs3091244) | CT | 88 (36) | 118(36) | T | 161 (33) | 196(31) | 0.451 | |
| TT | 29 (12) | 30(9) | A | 36(7) | 52(8) | 0.789 | ||
| CA | 17(7) | 28(8) | ||||||
| TA | 15(6) | 18(5) | ||||||
| AA | 2(1) | 3(1) | ||||||
| +90 | AA | 117(48.75) | 159(47.89) | 0.973 | A | 335 (69.8) | 461 (69.4) | 0.895 |
| (rs 1417938) | AT | 101 (42.08) | 143(43.07) | T | 145 (30.2) | 203(30.6) | ||
| TT | 22(9.17) | 30(9.04) | ||||||
| +838 | GG | 283 (83.98) | 395(88.17) | 0.125 | G | 619(91.8) | 840(93.8) | 0.14 |
| (rs 1800947) | GC | 53(15.73) | 50(11.16) | C | 55 (8.2) | 56(6.3) | ||
| CC | 1 (0.3) | 3 (0.67) | ||||||
| +2043 | GG | 142(42.51) | 207(46.31) | 0.538 | G | 441 (66) | 607(67.9) | 0.434 |
| (rs1205) | GA | 157(47.01) | 193(43.18) | A | 227 (34) | 287(32.1) | ||
| AA | 35(10.48) | 47(10.51) |
Association of CRP Haplotype with SLE Risk
We conducted pair-wise linkage disequilibrium (LD) analysis using four SNPs (excluding the tri-allelic SNP -390), and found different patterns of LD association in cases versus controls. In cases, with the exception of the -861/+90 and +90/+838 pairs, all SNP pairs are in significant LD. Among controls, all pairs were in significant LD except for the −861/+838 pair (Table 2). Because SNP haplotype may be more informative analyses of multiple SNP and the LD pattern may differ between cases and controls, we assessed the distribution of CRP haplotypes between cases and controls. All SNPs (-861, -390, +90, +838, +2043) were included in our global haplotype analysis using 222 cases and 313 controls for which 100% genotyping data were available (Table 3). A total of 8 haplotypes were observed at a frequency of 2% or greater from either case or control groups. Even accounting for two tests (the logistic regression and haplotype analyses) of the relationship between the CRP SNPs and case/control status, the overall haplotype distribution was significantly different between cases and controls (χ2 = 138.86, p < 0.000001) (Table 3). Haplotype 5 appears to be the most pronounced risk haplotype for SLE while haplotypes 2, 4 and 8 seem to convey protection against SLE. However, since no single allele at any locus defined and was restricted to a given risk or protective overall haplotype, no specific haplotype-tagging SNP could be identified to account for the significant overall haplotype associations.
Table 2.
Pairwise Linkage Disequilibrium between CRP SNPs
| Pairwise Linkage Disequilibrium - SLE | ||||
| +90 | +838 | +2043 | ||
| −861 | 0.018 (0.945) | 0.993 (0.037) | 0.880 (< 0.001) | |
| +90 | 0.066 (0.511) | 0.234 (0.016) | D' (p-value) | |
| +838 | 0.846 (< 0.001) | |||
| Pairwise Linkage Disequilibrium - Controls | ||||
| +90 | +838 | +2043 | ||
| −861 | 0.996 (< 0.001) | 0.482 (0.319) | 0.996 (< 0.001) | |
| +90 | 0.687 (0.002) | 0.952 (< 0.001) | D' (p-value) | |
| +838 | 0.753 (< 0.001) | |||
Table 3.
CRP Haplotype Case-Control Comparison
| Haplotype | −861 (T>C) | −390 (C>T>A) | +90 (A>T) | +838 (G>C) | +2043 (G>A) | SLE Frequency (n = 222) | Control Frequency (n = 313) | Frequency Difference |
|---|---|---|---|---|---|---|---|---|
| H1 | T | C | A | G | G | 0.330 | 0.304 | 0.026 |
| H2 | T | C | A | G | A | 0.183 | 0.259 | −0.076 |
| H3 | T | C | A | C | A | 0.047 | 0.050 | −0.004 |
| H4 | T | T | T | G | G | 0.201 | 0.286 | −0.085 |
| H5 | T | T | T | G | A | 0.061 | 0.002 | 0.058 |
| H6 | T | A | A | G | G | 0.038 | 0.011 | 0.027 |
| H7 | C | C | A | G | G | 0.032 | 0.004 | 0.028 |
| H8 | C | A | A | G | G | 0.007 | 0.061 | −0.054 |
| Overall p < 0.000001 | ||||||||
| χ2 = 138.86 | ||||||||
Association of CRP SNPs with SLE Clinical Characteristics
Because we detected a significant association between CRP SNP haplotypes and SLE, we perform follow-up analyses to determine if the CRP SNPs were associated with specific SLE characteristics. These tests were done in a subgroup of 237 SLE patients on whom we have clinical phenotype data, and we performed either ANOVA for quantitative clinical characteristics of SLE or logistic regression analysis for categorical variables adjusting for the effects of age, BMI, CRP levels, and smoking when appropriate. Individuals with +838 GC genotype also exhibited nominally significantly higher C4 levels compared to GG individuals (23.46 ± 8.35 vs. 20.67 ± 7.68, p = 0.033). No significant associations were observed between any of the individual SNPs and SLAM, SLICC, C3, creatinine, renal disease, arthritis, and antiphospholipid antibodies (data not shown).
CRP SNPs Associations with Serum C-Reactive Protein Levels
We performed both single-site and haplotype analyses to assess the association between the five CRP SNPs and log-transformed serum CRP levels (logCRP) in a subgroup of SLE patients (n = 237) in which CRP levels were available. After performing 5 tests of the single-site analyses, minor alleles of two SNPs revealed significant associations with increased logCRP in SLE patients (+90, p = 0.0032; -390, p = 0.012), even at the conservative Bonferroni level of significance (p = 0.01). Homozygotes of the less common allele (T) at +90 had the highest logCRP level (1.544 ± 0.271) compared to homozygotes of the wild type allele (0.639 ± 0.120) and the heterozygotes (0.623 ± 0.117). Mean logCRP levels were significantly higher in homozygotes of T allele at the triallelic promoter SNP-390 (1.305 ± 0.259) and heterozygotes with an A allele (CA) (1.356 ± 0.286) when compared to homozygotes of the wild type (CC) (0.519 ± 0.134). SNP -861 and the two SNPs (+838, +2043), which were previously found to be associated with decreased CRP levels by Russell et al. , did not show a statistically significant impact on circulating CRP levels in this cohort (Table 4).
Table 4.
Association of CRP Polymorphisms and Mean logCRP level (+/− SE)
| SNP | Genotype | n (%) | Mean ± S.E. | p-value |
|---|---|---|---|---|
| −861* | TT | 191 (84.14) | 0.718 ± 0.073 | 0.16 |
| (rs3093059) | TC | 36(15.86) | 0.977 ± 0.184 | |
| CC | 0(0) | ... | ||
| −390* | CC | 50 (34.48) | 0.519 ± 0.134 | 0.012 |
| (rs3091244) | CT | 55 (37.93) | 0.589 ± 0.121 | |
| TT | 19(13.10) | 1.305 ± 0.259 | ||
| CA | 11(7.59) | 1.356 ± 0.286 | ||
| TA | 8 (5.52) | 0.696 ± 0.351 | ||
| AA | 2(1.38) | 0.515 ± 0.596 | ||
| +90* | AA | 66 (45.83) | 0.639 ± 0.120 | 0.0032 |
| (rs1417938) | AT | 63 (43.75) | 0.623 ± 0.117 | |
| TT | 15 (10.42) | 1.544 ± 0.271 | ||
| +838* | GG | 193 (83.91) | 0.703 ± 0.075 | 0.373 |
| (rs 1800947) | GC | 37 (16.09) | 0.869 ± 0.161 | |
| CC | 0(0) | ... | ||
| +2043* | GG | 100 (43.67) | 0.845 ± 0.112 | 0.207 |
| (rs1205) | GA | 107 (46.72) | 0.605 ± 0.092 | |
| AA | 22 (9.61) | 0.857 ± 0.213 |
Mean logCRP level (± SE) - Adjusted for Age, BMI, and Smoking
CRP Haplotype Association with Serum CRP Levels
Given the significant individual effects of SNP -390, and +90 have on CRP levels and the LD between these SNPs, and the observation that +90 is in high LD with -861, whereas -861 is in high LD with +838 and +2043 in cases, we performed 3-SNP haplotype analysis consisting of these three potentially functional SNP to evaluate the significance of the CRP promoter region has on CRP levels. Three-SNP haplotypes were inferred using the haplo.glm function in the haplo.stats package in R. Haplotype -861C X -390T X +90T (H3 in Table 5) was associated with an increase of 1.171 logCRP units compared to the reference haplotype ( p = 0.0161), and is consistent with the results in Table 4 (the individual SNP associations). Haplotype -861T X -390T X +90T also associated with an increase of logCRP by 0.2928 ( p = 0.0423) (H6 in Table 5).
Table 5.
Association of CRP Promoter Haplotype with Serum logCRP Levels in SLE
| 3-Loci Promoter Haplotype | ||||||||
|---|---|---|---|---|---|---|---|---|
| −861 (T>C) | −390 (C>T>A) | +90 (A>T) | Haplotype Frequency | Coefficient | SE | t-stat | p-value | |
| NOTE. -- tstatistics and p values were calculated from the coefficients and SEs within the best-fit multivariate model by the haplo.glm function in the haplo.stats R package. | ||||||||
| Intercept | ... | ... | ... | ... | −1.651 | 0.364 | −4.536 | 0.000 |
| Age | ... | ... | ... | ... | 0.014 | 0.007 | 2.064 | 0.040 |
| BMI | ... | ... | ... | ... | 0.048 | 0.010 | 4.887 | 0.000 |
| Smoke | ... | ... | ... | ... | 0.394 | 0.137 | 2.878 | 0.004 |
| H2 | C | C | A | 0.050 | 0.028 | 0.273 | 0.103 | 0.918 |
| H3 | C | T | T | 0.019 | 1.171 | 0.483 | 2.425 | 0.016 |
| H4 | T | A | A | 0.074 | 0.263 | 0.246 | 1.072 | 0.285 |
| H5 | T | T | A | 0.024 | −0.158 | 0.384 | −0.411 | 0.682 |
| H6 | T | T | T | 0.301 | 0.293 | 0.143 | 2.042 | 0.042 |
| H_other* | * | * | * | 0.009 | 0.648 | 0.682 | 0.950 | 0.343 |
| H1 = Referent | T | C | A | 0.522 | Referent | ... | ... | ... |
Haplotypes with frequency <2% were pooled as “H_other.”
DISCUSSION
We examined the association of CRP tagSNPs in relation to SLE risk and CRP levels in SLE patients. Individually none of the examined SNPs showed significant association with SLE risk. Previously, Russell et al.39 reported that minor (A) allele of +2043 was associated with SLE risk; but this association is not confirmed in our sample. However, in contrast to the single-site analysis, the CRP haplotype analyses yielded significant associations with SLE risk. The global 5-site CRP haplotype distribution was significantly different (nominal p < 0.000001), between cases and controls even after conservative Bonferroni adjustment for multiple comparisons (critical p-value = 0.05/5 tests = 0.01). Further inspection of the haplotype results indicated that no single haplotype-tagging SNP explained the significant haplotype association with SLE risk, consistent with the individual SNP results.
The observed significant haplotype association in the absence of individual SNP association may be explained in several ways. First, the use of multilocus analyses in the SNP setting should improve the information content of genomic regions40] and may also capture effects from multiple polymorphisms (versus single tag SNP analysis) as well as subtle interaction effects (epistasis)[41] within the given haplotype block. Second, haplotypes can mark unique chromosomal segments that harbor susceptibility alleles, even if the linkage disequilibrium (LD) patterns differ between study populations. Given that LD patterns are likely to differ between groups due to population history and/or genetic admixture, our finding that CRP haplotype differ between SLE cases and controls is consistent with previous results with Russell et al.39. Both studies indicate that CRP polymorphisms are associated with SLE risk; however, neither study has identified strong evidence of a specific susceptibility allele. Even though the individual tag SNP approach has been the gold standard for association studies for many years, it requires that a SNP be in strong LD with a causal polymorphism that has a measurable effect. Given the polygenic and multifactorial nature of SLE pathogenesis, the haplotype approach may be more useful in detecting genotype-phenotype associations in comparison to the individual SNP approach, especially if multiple and/or uncommon variants are associated with the SLE risk. As noted in the methods, there are 31 SNPs reported in the SeattleSNPs Program for Genomic Applications, of which five were sufficiently polymorphic for analysis purposes in our study sample.
Genetic variation at the CRP locus could influence SLE risk via its effect on CRP levels and Russell et al.[39 reported a significant association between CRP levels and genotypes at SNPs +2043 and +838. We did not observe significant association between CRP levels and genotypes at theses two SNPs, but we did observe significant association between genotypes at a promoter SNP (SNP-390, nominal p=0.01) and a SNP at an intron/exon boundary (SNP+90, nominal p=0.003), even after Bonferroni adjustment for five tests. In fact, our result that the -390T allele is associated with increased CRP levels directly supports previous observations that T allele forms an E-Box binding cite, which is involved in transcription binding[42]. Additional in silico prediction analysis and in vitro data confirmed that haplotypes containing the -390T allele increases reporter gene activity significantly[43]. Among our SLE cases, homozygotes for the -390T allele had a two-fold increase in logCRP (1.544 ± 0.271) compared to homozygotes of the wild type allele (0.639 ± 0.120) and heterozygotes (0.623 ± 0.117, p = 0.0032) (Table 4).
Polymorphisms located in gene promoters may play a role in gene function by altering transcription factor identification and binding, which in turn can influence gene expression and affect biological pathways. Likewise, SNPs at the intron/exon boundary may result in alternative splicing and affect gene function. In fact, these relationships between CRP levels at these two individual SNPs have also been reported in healthy, non-SLE populations[43-45]. In addition, our haplotype analyses of the 2 promoter and 1 splice site SNPs (Table 5) further indicate that all three SNPs contribute to increased CRP levels.
Previously Russell et al.[29] reported significant association between SNPs +838, +2043 and decreased CRP level in a British SLE cohort. Similarly, a more recent study by Miller et al.[44] reported the same association of these two SNPs in three large cohorts of healthy general population. However, we did not observe the same association between SNPs +838, +2043 and decreased CRP in our 273 SLE patients. The lack of association in our SLE women may be attributed to the limited sample size of the minor allele carriers in our study, or it may be confounded by the effects from anti-inflammatory medications SLE patients take on the regular basis, like corticosteroids.
Although our association results between CRP genotypes and CRP levels are consistent with some previous reports, they differ from others. Determining the true association between genetic variation and CRP levels is inherently difficult due to the complex mechanism of CRP production, which is activated by cytokines IL-6 and IL-1 and influenced by multiple other genes and environmental factors[46]. The difficulty is compounded in SLE cohort because not only are the inflammatory cytokines increased in SLE patients[47], but the strong correlation between CRP and IL-6 levels in healthy subjects may be absent in SLE[48]. SLE is a chronic inflammatory disease with abnormal expression of CRP during both the presence and absence of acute infections. Multiple studies have also found inconsistent correlations between CRP levels and SLE disease activity, indicating that the mechanism influencing CRP expression in SLE individuals may differ from that in the general population. Our data showed that even though the exon 2 and 3' regions SNPs did not correlate with significantly decreased CRP levels in our SLE patients as others have found, individual SNPs and haplotype in the promoter region revealed associations with increased basal CRP levels as shown in the general population. Our findings emphasize the important functional role of CRP promoter polymorphisms may play in their expression even in patients with a chronic inflammatory disease.
Russell et al. 's family-based study proposed that low basal levels of CRP may predispose to antinuclear autoantibody production, which in turn contributes to the development of human lupus39. Our results show that individually, certain SNPs are correlated with CRP levels, but their association with SLE risk was not significant. Although we did not find strong evidence that any of the individual CRP SNPs influence CRP level and thereby predict SLE risk, the significant global haplotype results suggest that variation in the CRP gene modifies SLE risk via as yet unidentified mechanisms. Our individual SNPs results, coupled with promoter haplotype results, confirm the previous studies done in the general population that CRP promoter variants have a significant impact on CRP levels in SLE patients. The lack of association with SNPs +838, +2043 and decreased CRP levels in our patients may result from years of SLE insult (mean disease duration of 10.13 years) from the chronic inflammatory state.
It remains a possibility that CRP itself does not directly contribute to SLE susceptibility, rather one or more as yet unidentified SLE susceptibility alleles in nearby loci may be in strong LD with one or more of the CRP SNPs we examined. Two potential SLE susceptibility genes that also mapped to 1q23, FcγRIIA and FcγRIIIA , encode for low-affinity receptors for IgG. Recent meta-analyses revealed that the FcγRIIA-R/H131 polymorphism was associated with a 1.3-fold greater risk of development of lupus, and that the FcγRIIIA -V/F158 polymorphism conferred 1.4-fold risk for developing lupus nephritis[49]. The interaction of IgG Fc receptors containing an activation motif (ITAM) with immune complexes and cytotoxic autoantibodies can initiate an inflammatory response leading to tissue damage[50]. It has also been demonstrated that FcγRIIA-R/H131, working in conjunction with CRP, has the unique ability to alter the cytokine profile of the host[51] by mediating phagocytosis[52], and contributing to the impaired removal of circulating immune complexes[53], resulting in the antibody-triggered inflammation and disease pathogenesis of SLE and nephritis. Given the overlapping chromosomal position of the human CRP, FcγRIIA and FcγRIIIA genes and their unique ability to modify SLE phenotype when working together, it is likely that genetic interaction between these three loci (epistasis) may modify SLE susceptibility.
In summary, our study is consistent with some previous studies that genetic variation in CRP influences risk of SLE and levels of CRP in SLE patients although the same genetic variation did not influence both CRP levels and SLE risk in our study and some of our single SNP association results differ from those of other studies. These results may indicate that the SNPs may not act via level alone, but exerts their effects via different kinds of activity or interactions with other proteins. Furthermore, complex diseases and traits (such as SLE and CRP levels) are likely to be influenced by multiple genes each exerting effects in small to modest range[54]. A limitation of our study is the relatively small sample size which reduces our ability to detect genes with small effects as well as effects of gene by gene and gene by environment interactions. In addition, as circulating CRP is a sensitive acute-phase protein that could easily be fluctuated by multiple factors, longitudinal studies of CRP and SLE would be useful. Future work is necessary, perhaps using murine models[27,28 to determine the true mechanism underlying the associations between CRP genetic variation and SLE risk.. Such information will further our understanding of SLE etiology and may have direct clinical relevance.
ACKNOWLEDGEMENTS
The authors thank Dr. Christopher Carlson from the University of Washington for his assistance in the triallelic SNP -390 genotyping protocol.
Supported in parts by the National Heart, Lung, and Blood Institute Grants HL 074165 and HL 54900, the National Institute of Arthritis and Musculoskeletal and Skin Diseases Grants AR 46588 and AR 002213, and a grant from the NIH General Clinical Research Center M01-RR000056.
REFERENCES
- 1.Kyttaris VC, Katsiari CG, Juang Y-T, Tsokos GC. New insights into the pathogenesis of systemic lupus erythematosus. Current Rheumatology Reports. 2005;7:469–475. doi: 10.1007/s11926-005-0054-3. [DOI] [PubMed] [Google Scholar]
- 2.Block SR, Winfield JB, Lockshin MD, D'Angelo WA, Christian CL. Studies of twins with systemic lupus erythematosus. A review of the literature and presentation of 12 additional sets. American Journal of Medicine. 1975;59:533–552. doi: 10.1016/0002-9343(75)90261-2. [DOI] [PubMed] [Google Scholar]
- 3.Alarcon-Segovia D, Alarcon-Riquelme ME, Cardiel MH, Caeiro F, Massardo L, Villa AR, Pons-Estel BA, Grupo Latinoamericano de Estudio del Lupus E Familial aggregation of systemic lupus erythematosus, rheumatoid arthritis, and other autoimmune diseases in 1,177 lupus patients from the gladel cohort. Arthritis Rheum. 2005;23(6):770–1. doi: 10.1002/art.20999. 52:1138-1147. [DOI] [PubMed] [Google Scholar]
- 4.Lawrence JS, Martins CL, Drake GL. A family survey of lupus erythematosus. 1. Heritability. Journal of Rheumatology. 1987;14:913–921. [PubMed] [Google Scholar]
- 5.Tsao BP. Update on human systemic lupus erythematosus genetics. Current Opinion in Rheumatology. 2004;16:513–521. doi: 10.1097/01.bor.0000132648.62680.81. [DOI] [PubMed] [Google Scholar]
- 6.Johanneson B, Lima G, von Salome J, Alarcon-Segovia D, Alarcon-Riquelme ME, Collaborative Group on the Genetics of Sle TBIICotGoSLEaSs A major susceptibility locus for systemic lupus erythemathosus maps to chromosome 1q31. American Journal of Human Genetics. 2002;71:1060–1071. doi: 10.1086/344289. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Mohan C, Alas E, Morel L, Yang P, Wakeland EK. Genetic dissection of sle pathogenesis. Sle1 on murine chromosome 1 leads to a selective loss of tolerance to h2a/h2b/DNA subnucleosomes. Journal of Clinical Investigation. 1998;101:1362–1372. doi: 10.1172/JCI728. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Moser KL, Neas BR, Salmon JE, Yu H, Gray-McGuire C, Asundi N, Bruner GR, Fox J, Kelly J, Henshall S, Bacino D, Dietz M, Hogue R, Koelsch G, Nightingale L, Shaver T, Abdou NI, Albert DA, Carson C, Petri M, Treadwell EL, James JA, Harley JB. Genome scan of human systemic lupus erythematosus: Evidence for linkage on chromosome 1q in african-american pedigrees. Proc Natl Acad Sci U S A. 1998;95:14869–14874. doi: 10.1073/pnas.95.25.14869. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tsao BP, Cantor RM, Grossman JM, Kim SK, Strong N, Lau CS, Chen C-J, Shen N, Ginzler EM, Goldstein R, Kalunian KC, Arnett FC, Wallace DJ, Hahn BH. Linkage and interaction of loci on 1q23 and 16q12 may contribute to susceptibility to systemic lupus erythematosus. Arthritis Rheum. 2002;23(6):770–1. doi: 10.1002/art.10590. 46:2928-2936. [DOI] [PubMed] [Google Scholar]
- 10.Scofield RH, Bruner GR, Kelly JA, Kilpatrick J, Bacino D, Nath SK, Harley JB. Thrombocytopenia identifies a severe familial phenotype of systemic lupus erythematosus and reveals genetic linkages at 1q22 and 11p13. Blood. 2003;101:992–997. doi: 10.1182/blood-2002-04-1003. [DOI] [PubMed] [Google Scholar]
- 11.Kushner I. The acute phase response: An overview. Methods in Enzymology. 1988;163:373–383. doi: 10.1016/0076-6879(88)63037-0. [DOI] [PubMed] [Google Scholar]
- 12.Robey FA, Jones KD, Tanaka T, Liu TY. Binding of c-reactive protein to chromatin and nucleosome core particles. A possible physiological role of c-reactive protein. Journal of Biological Chemistry. 1984;259:7311–7316. [PubMed] [Google Scholar]
- 13.Du Clos TW, Zlock LT, Rubin RL. Analysis of the binding of c-reactive protein to histones and chromatin. Journal of Immunology. 1988;141:4266–4270. [PubMed] [Google Scholar]
- 14.Gershov D, Kim S, Brot N, Elkon KB. C-reactive protein binds to apoptotic cells, protects the cells from assembly of the terminal complement components, and sustains an antiinflammatory innate immune response: Implications for systemic autoimmunity. Journal of Experimental Medicine. 2000;192:1353–1364. doi: 10.1084/jem.192.9.1353. [erratum appears in j exp med 2001 jun 18;193(12):1439] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Szalai AJ. C-reactive protein (crp) and autoimmune disease: Facts and conjectures. Clinical & Developmental Immunology. 2004;11:221–226. doi: 10.1080/17402520400001751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kravitz MS, Pitashny M, Shoenfeld Y. Protective molecules--c-reactive protein (crp), serum amyloid p (sap), pentraxin3 (ptx3), mannose-binding lectin (mbl), and apolipoprotein a1 (apo a1), and their autoantibodies: Prevalence and clinical significance in autoimmunity. Journal of Clinical Immunology. 2005 Nov;25(6):582–91. doi: 10.1007/s10875-005-7828-2. [DOI] [PubMed] [Google Scholar]
- 17.Szalai AJ, Weaver CT, McCrory MA, van Ginkel FW, Reiman RM, Kearney JF, Marion TN, Volanakis JE. Delayed lupus onset in (nzb x nzw)f1 mice expressing a human c-reactive protein transgene. Arthritis Rheum. 2003;23(6):770–1. doi: 10.1002/art.11026. [see comment] 48:1602-1611. [DOI] [PubMed] [Google Scholar]
- 18.Rodriguez W, Mold C, Marnell LL, Hutt J, Silverman GJ, Tran D, Du Clos TW. Prevention and reversal of nephritis in mrl/lpr mice with a single injection of c-reactive protein. Arthritis Rheum. 2006;23(6):770–1. doi: 10.1002/art.21556. 54:325-335. [DOI] [PubMed] [Google Scholar]
- 19.Maury CP, Wegelius O. Clinical value of serum amyloid a and c-reactive protein measurements in secondary amyloidosis. International Journal of Tissue Reactions. 1985;7:405–407. [PubMed] [Google Scholar]
- 20.Becker GJ, Waldburger M, Hughes GR, Pepys MB. Value of serum c-reactive protein measurement in the investigation of fever in systemic lupus erythematosus. Annals of the Rheumatic Diseases. 1980;39:50–52. doi: 10.1136/ard.39.1.50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Morrow WJ, Isenberg DA, Parry HF, Snaith ML. C-reactive protein in sera from patients with systemic lupus erythematosus. Journal of Rheumatology. 1981;8:599–604. [PubMed] [Google Scholar]
- 22.Eberhard OK, Haubitz M, Brunkhorst FM, Kliem V, Koch KM, Brunkhorst R. Usefulness of procalcitonin for differentiation between activity of systemic autoimmune disease (systemic lupus erythematosus/systemic antineutrophil cytoplasmic antibody-associated vasculitis) and invasive bacterial infection. Arthritis Rheum. 1997;23(6):770–1. doi: 10.1002/1529-0131(199707)40:7<1250::AID-ART9>3.0.CO;2-A. [see comment] 40:1250-1256. [DOI] [PubMed] [Google Scholar]
- 23.Roy S, Tan KT. Pyrexia and normal c-reactive protein (crp) in patients with systemic lupus erythematosus: Always consider the possibility of infection in febrile patients with systemic lupus erythematosus regardless of crp levels. Rheumatology. 2001;40:349–350. doi: 10.1093/rheumatology/40.3.349. [DOI] [PubMed] [Google Scholar]
- 24.Inoue T, Kanayama Y, Katoh N, Ohe A, Amatsu K, Okamura M, Horiguchi T. Significance of c-reactive protein elevation in the pretreatment stage of systemic lupus erythematosus. Scand J Rheumatol. 1981;10:222–224. doi: 10.3109/03009748109095302. [DOI] [PubMed] [Google Scholar]
- 25.Linares LF, Gomez-Reino JJ, Carreira PE, Morillas L, Ibero I. C-reactive protein (crp) levels in systemic lupus erythematosus (sle) Clin Rheumatol. 1986;5:66–69. doi: 10.1007/BF02030970. [DOI] [PubMed] [Google Scholar]
- 26.Hind CR, Ng SC, Feng PH, Pepys MB. Serum c-reactive protein measurement in the detection of intercurrent infection in oriental patients with systemic lupus erythematosus. Annals of the Rheumatic Diseases. 1985;44:260–261. doi: 10.1136/ard.44.4.260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bertouch JV, Roberts-Thompson PJ, Feng PH, Bradley J. C-reactive protein and serological indices of disease activity in systemic lupus erythematosus. Annals of the Rheumatic Diseases. 1983;42:655–658. doi: 10.1136/ard.42.6.655. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Barnes EV, Narain S, Naranjo A, Shuster J, Segal MS, Sobel ES, Armstrong AE, Santiago BE, Reeves WH, Richards HB. High sensitivity c-reactive protein in systemic lupus erythematosus: Relation to disease activity, clinical presentation and implications for cardiovascular risk. Lupus. 2005;14:576–582. doi: 10.1191/0961203305lu2157oa. [DOI] [PubMed] [Google Scholar]
- 29.Russell AI, Cunninghame Graham DS, Shepherd C, Roberton CA, Whittaker J, Meeks J, Powell RJ, Isenberg DA, Walport MJ, Vyse TJ. Polymorphism at the c-reactive protein locus influences gene expression and predisposes to systemic lupus erythematosus. Human Molecular Genetics. 2004;13:137–147. doi: 10.1093/hmg/ddh021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield NF, Schaller JG, Talal N, Winchester RJ. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum. 1982;23(6):770–1. doi: 10.1002/art.1780251101. 25:1271-1277. [DOI] [PubMed] [Google Scholar]
- 31.Hochberg MC. Updating the american college of rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum. 1997;23(6):770–1. doi: 10.1002/art.1780400928. [see comment] 40:1725. [DOI] [PubMed] [Google Scholar]
- 32.Selzer F, Sutton-Tyrrell K, Fitzgerald S, Tracy R, Kuller L, Manzi S. Vascular stiffness in women with systemic lupus erythematosus. Hypertension. 2001;37:1075–1082. doi: 10.1161/01.hyp.37.4.1075. [DOI] [PubMed] [Google Scholar]
- 33.Macy EM, Hayes TE, Tracy RP. Variability in the measurement of c-reactive protein in healthy subjects: Implications for reference intervals and epidemiological applications. Clinical Chemistry. 1997;43:52–58. [PubMed] [Google Scholar]
- 34.Cushman M, Cornell ES, Howard PR, Bovill EG, Tracy RP. Laboratory methods and quality assurance in the cardiovascular health study. Clinical Chemistry. 1995;41:264–270. [PubMed] [Google Scholar]
- 35.Liang MH, Socher SA, Larson MG, Schur PH. Reliability and validity of six systems for the clinical assessment of disease activity in systemic lupus erythematosus. Arthritis Rheum. 1989;23(6):770–1. doi: 10.1002/anr.1780320909. 32:1107-1118. [DOI] [PubMed] [Google Scholar]
- 36.Gladman D, Ginzler E, Goldsmith C, Fortin P, Liang M, Urowitz M, Bacon P, Bombardieri S, Hanly J, Hay E, Isenberg D, Jones J, Kalunian K, Maddison P, Nived O, Petri M, Richter M, Sanchez-Guerrero J, Snaith M, Sturfelt G, Symmons D, Zoma A. The development and initial validation of the systemic lupus international collaborating clinics/american college of rheumatology damage index for systemic lupus erythematosus. Arthritis Rheum. 1996;23(6):770–1. doi: 10.1002/art.1780390303. [see comment] 39:363-369. [DOI] [PubMed] [Google Scholar]
- 37.Lewontin RC. The interaction of selection and linkage. Ii. Optimum models. Genetics. 1964 Oct;50:757–82. doi: 10.1093/genetics/50.4.757. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Terwilliger JD, Ott. J. Johns Hopkins University Press; Baltimore: 1994. Handbook of human genetic linkage. [Google Scholar]
- 39.Lake SL, Lyon H, Tantisira K, Silverman EK, Weiss ST, Laird NM, Schaid DJ. Estimation and tests of haplotype-environment interaction when linkage phase is ambiguous. Human Heredity. 2003;55:56–65. doi: 10.1159/000071811. [DOI] [PubMed] [Google Scholar]
- 40.Ott J, Rabinowitz D. The effect of marker heterozygosity on the power to detect linkage disequilibrium. Genetics. 1997;147:927–930. doi: 10.1093/genetics/147.2.927. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Davidson S. Research suggests importance of haplotypes over snps. Nat biotechnol. 2000;18:1134–1135. doi: 10.1038/81100. [DOI] [PubMed] [Google Scholar]
- 42.Szalai AJ, Wu J, Lange EM, McCrory MA, Langefeld CD, Williams A, Zakharkin SO, George V, Allison DB, Cooper GS, Xie F, Fan Z, Edberg JC, Kimberly RP. Single-nucleotide polymorphisms in the c-reactive protein (crp) gene promoter that affect transcription factor binding, alter transcriptional activity, and associate with differences in baseline serum crp level. Journal of Molecular Medicine. 2005;83:440–447. doi: 10.1007/s00109-005-0658-0. [see comment] [DOI] [PubMed] [Google Scholar]
- 43.Carlson CS, Aldred SF, Lee PK, Tracy RP, Schwartz SM, Rieder M, Liu K, Williams OD, Iribarren C, Lewis EC, Fornage M, Boerwinkle E, Gross M, Jaquish C, Nickerson DA, Myers RM, Siscovick DS, Reiner AP. Polymorphisms within the c-reactive protein (crp) promoter region are associated with plasma crp levels. Am J Hum Genet. 2005;77:64–77. doi: 10.1086/431366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Miller DT, Zee RYL, Suk Danik J, Kozlowski P, Chasman DI, Lazarus R, Cook NR, Ridker PM, Kwiatkowski DJ. Association of common crp gene variants with crp levels and cardiovascular events. Ann Hum Genet. 2005;69:623–638. doi: 10.1111/j.1529-8817.2005.00210.x. [DOI] [PubMed] [Google Scholar]
- 45.Lange LA, Carlson CS, Hindorff LA, Lange EM, Walston J, Durda JP, Cushman M, Bis JC, Zeng D, Lin D, Kuller LH, Nickerson DA, Psaty BM, Tracy RP, Reiner AP. Association of polymorphisms in the crp gene with circulating c-reactive protein levels and cardiovascular events. JAMA. 2006 Dec 13;296(22):2703–11. doi: 10.1001/jama.296.22.2703. [DOI] [PubMed] [Google Scholar]
- 46.Wessel J, Moratorio G, Rao F, Mahata M, Zhang L, Greene W, Rana BK, Kennedy B, Khandrika S, Huang P, Lillie EO, Shih PB, Smith DW, Wen G, Hamilton BA, Ziegler MG, Witztum JL, Schork NJ, Schmid-Schönbein GW, O'Connor DT. C reactive protein, an “intermediate phenotype” for inflammation: Human twin studies reveal heritability, association with blood pressure and the metabolic syndrome, and the influence of common polymorphism at catecholaminergic/β-adrenergic pathway loci. J Hypertension. 2007;25:329–343. doi: 10.1097/HJH.0b013e328011753e. [DOI] [PubMed] [Google Scholar]
- 47.Asanuma Y, Chung CP, Oeser A, Shintani A, Stanley E, Raggi P, Stein CM. Increased concentration of proatherogenic inflammatory cytokines in systemic lupus erythematosus: Relationship to cardiovascular risk factors. Journal of Rheumatology. 2006 Mar;33(3):539–45. [PubMed] [Google Scholar]
- 48.Williams RC, Jr., Harmon ME, Burlingame R, Du Clos TW. Studies of serum c-reactive protein in systemic lupus erythematosus. J Rheumatol. 2005;32:454–461. [PubMed] [Google Scholar]
- 49.Karassa FB, Trikalinos TA, Ioannidis JPA. The role of fcgammariia and iiia polymorphisms in autoimmune diseases. Biomedicine & Pharmacotherapy. 2004;58:286–291. doi: 10.1016/j.biopha.2004.04.004. [DOI] [PubMed] [Google Scholar]
- 50.Clynes R, Calvani N, Croker BP, Richards HB. Modulation of the immune response in pristane-induced lupus by expression of activation and inhibitory fc receptors. Clin Exp Immunol. 2005;141:230–237. doi: 10.1111/j.1365-2249.2005.02847.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Mold C, Rodriguez W, Rodic-Polic B, Du Clos TW. C-reactive protein mediates protection from lipopolysaccharide through interactions with fc gamma r. J Immunol. 2002;169:7019–7025. doi: 10.4049/jimmunol.169.12.7019. [DOI] [PubMed] [Google Scholar]
- 52.Bodman-Smith KB, Gregory RE, Harrison PT, Raynes JG. Fcgammariia expression with fcgammari results in c-reactive protein- and igg-mediated phagocytosis. J Leukoc Biol. 2004;75:1029–1035. doi: 10.1189/jlb.0703306. [DOI] [PubMed] [Google Scholar]
- 53.Zuniga R, Markowitz GS, Arkachaisri T, Imperatore EA, D'Agati VD, Salmon JE. Identification of igg subclasses and c-reactive protein in lupus nephritis: The relationship between the composition of immune deposits and fcgamma receptor type iia alleles. Arthritis Rheum. 2003;48:460–470. doi: 10.1002/art.10930. [DOI] [PubMed] [Google Scholar]
- 54.Consortium. WTCC Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007;447:661–678. doi: 10.1038/nature05911. [DOI] [PMC free article] [PubMed] [Google Scholar]