The ITGAV rs3738919 variant and susceptibility to rheumatoid arthritis in four Caucasian sample sets

Introduction Angiogenesis is an important process in the development of destructive synovial pannus in rheumatoid arthritis (RA). The ITGAV +gene encodes a cell cycle-associated antigen, integrin ανβ 3, which plays a role in RA angiogenesis. Previously, two independent studies identified an association between the major allele of the ITGAV single-nucleotide polymorphism (SNP) rs3738919 and RA. We therefore tested this association in an independent study using New Zealand (NZ) and Oxford (UK) RA case control samples. Methods We compared genotype frequencies in 740 NZ Caucasian RA patients and 553 controls genotyped for rs3738919, using a polymerase chain reaction-restriction fragment length polymorphism assay. A TaqMan genotyping SNP assay was used to type 713 Caucasian RA patients and 515 control samples from Oxford for the rs3738919 variant. Association of rs3738919 with RA was tested in these two sample sets using the chi-square goodness-of-fit test. The Mantel-Haenszel test was used to perform a meta-analysis, combining the genetic results from four independent Caucasian case control cohorts, consisting of 3,527 cases and 4,126 controls. Haplotype analysis was also performed using SNPs rs3911238, rs10174098 and rs3738919 in the Wellcome Trust Case Control Consortium, NZ and Oxford case control samples. Results We found no evidence for association between ITGAV and RA in either the NZ or Oxford sample set (odds ratio [OR] = 0.88, Pallelic = 0.11 and OR = 1.18, Pallelic = 0.07, respectively). Inclusion of these data in a meta-analysis (random effects) of four independent cohorts (3,527 cases and 4,126 controls) weakens support for the hypothesis that rs3738919 plays a role in the development of RA (ORcombined = 0.92, 95% confidence interval 0.80 to 1.07; P = 0.29). No consistent haplotype associations were evident. Conclusions Association of ITGAV SNP rs7378919 with RA was not replicated in NZ or Oxford case control sample sets. Meta-analysis of these and previously published data lends limited support for a role for the ITGAV in RA in Caucasians of European ancestry.


Introduction
Rheumatoid arthritis (RA) is a common systemic autoimmune disease characterised by chronic synovial inflammation leading to the formation of invasive, destructive pannus. The extensive formation of new blood vessels within the affected joint (hyperangiogenesis) is an important component of pannus formation [1].
The integrin ανβ3 (ITGAV) locus contains 30 exons spanning more than 90 kb of genomic DNA on human chromosome 2q31. It encodes the αν subunit of the cell cycle-associated antigen, integrin ανβ3, which plays a major role in RA angiogenesis. Angiogenesis is stimulated in RA by the increased metabolic demand of the pathologically active synovial tissues [1,[14][15][16]. Potentially, inhibition of this angiogenesis might suppress the destructive activities of pannus and even control disease activity [17]. This is supported by studies in animal models in which injection of ανβ3 antagonists has shown inhibition of neovascularisation and attenuation of joint inflammation [18].
A previous genome-wide linkage scan suggested 19 non-HLA regions contributing to RA in a French population [19]. One of these regions, on human chr2q31, contains the ITGAV gene (CD51). Subsequently, Jacq and colleagues [20] demonstrated an association between RA and the ITGAV rs3738919 C allele in a French Caucasian population (odds ratio for allele frequency difference [OR allelic ] = 0.77, 95% confidence interval [CI] 0.63 to 0.94). Significant association with the rs3738919 C allele was also demonstrated from imputed data by the Wellcome Trust Case Control Consortium (WTCCC) (OR allelic = 0.91, 95% CI 0.83 to 1.00) [21]. Collectively, these studies suggest a role for ITGAV in RA and justify further investigation of this locus. We therefore tested this association in an independent study using New Zealand (NZ) and Oxford (UK) RA case control samples.

Study subjects
The NZ population-based Caucasian sample consists of 740 RA patients fulfilling the American College of Rheumatology (ACR) criteria for RA [22]. Of the patients for whom data were available, 34.3% (234/683) were male, 82.9% (538/649) were rheumatoid factor (RF)-positive, 68.1% (275/404) were anti-cyclic citrullinated peptide (anti-CCP)-positive and 79.4% (576/725) carried the HLA-DRB1 shared epitope. Ethical approval for recruitment of cases was given by the New Zealand Multi-Region Ethics Committee, and recruitment of the controls was approved by the Lower South Ethics Committee. All patients provided written informed consent for the collection of samples and subsequent analysis. The control sample consisted of 553 NZ European Caucasians (226/552 male; 40.9%) with no history of autoimmune disease.
The 713 UK patients were recruited in Oxford, with informed consent from attendees at the rheumatology outpatient clinic at the Nuffield Orthopaedic Centre. All fulfilled the 1987 ACR criteria for RA; the average age of onset was 48 years, 28% were male, 77% were positive for RF and 77% carried the HLA-DRB1 shared epitope. Healthy ethnically matched controls (n = 515) were recruited from the same locale as that of blood donors. Approval for the study was given by the Oxford Research Ethics Committee (OxRec number C02.032).

DNA extraction and genotyping
DNA was extracted from peripheral blood samples of the RA patients and controls by means of guanidine isothiocyanatechloroform extraction methods. NZ study participants were genotyped for the ITGAV single-nucleotide polymorphism (SNP) rs3738919 using a polymerase chain reaction-restriction fragment length polymorphic SNP genotyping assay as follows: forward primer CACTTTCTGTAAATTAGTGTTA-GATCAAAAGG and reverse primer GCTTATAACTCACAAT-TCAGATTTTTGCC (primers from Sigma-Genosys, Sydney, Australia). The C allele (major allele) of the rs3738919 product (286 base pairs) was digested using the AluI restriction enzyme to form fragments of 223 and 63 base pairs. Oxford study participants were genotyped for rs3738919 using the TaqMan genotyping assay C___1278131_1, and both the NZ and Oxford samples were genotyped for the rs10174098 and rs3911238 ITGAV variants using the TaqMan genotyping assays C__30567648_10 and C___7617051_10, respectively, from Applied Biosystems (Scoresby, Australia).
Although the imputed rs3738919 data were available and were previously reported by Ahnert and Kirsten [21], we reimputed rs3738919 genotypes in order to provide information on imputation parameters. RA case and control genotypes were imputed from the WTCCC dataset using IMPUTE software [23]. Genotypes were imputed from 89% of cases (n = 1,659) and 90% of controls (n = 2,639) using a 7-Mb region and a calling threshold of 0.7.

Statistical analysis
An a priori power calculation was made based on the combined French [20] and WTCCC [21] OR (OR allelic = 1.16), using 800 cases and 600 controls with a major allele frequency of 65%. The power to detect association of (page number not for citation purposes) rs3738919 to RA using either the NZ or Oxford sample set was estimated to be 47%, using α = 0.05. ITGAV SNPs were tested for deviation from Hardy-Weinberg equilibrium in both the control and RA samples using a chi-square goodness-offit test. The significance of differences in the minor allele frequency between RA patients and controls, and stratified patients, was assessed using the chi-square goodness-of-fit test.
Of the five ITGAV SNPs (rs3911238, rs2887827, rs10174098, rs13006571 and rs16828163) genotyped by the WTCCC and rs3738919, only three (rs3911238, rs10174098 and rs3738919) were required to tag all major haplotypes defined by the six variants. Haplotype association analysis was performed using the SHEsis software package [24], which uses a full-precise-iteration algorithm to construct haplotypes and estimate haplotype frequencies.
Meta-analysis combining the French [20], WTCCC [25], NZ and Oxford samples was performed using STATA version 8.0 (StataCorp LP, College Station, TX, USA). The Mantel-Haen-szel test was used to estimate the average conditional common OR between the four independent sample sets and to test for any heterogeneity between the four groups using both fixed and random effects. R software [26] was used to perform the Cochran-Armitage trend test to determine recessive, additive and dominant trend values (Table 1).  [20] and WTCCC [21] data, with the minor allele under-represented in the case groups in all three cohorts (Table 1). However, the allele distribution in the Oxford RA cohort differs in that the minor allele is over-represented in the patient group compared with the control group (Table 1).

Haplotype analysis
Association analysis using our imputed genotypes for rs3738919 (P = 0.04) from the WTCCC data were consistent with the previous report of an association of this SNP with RA in the WTCCC [21]. In our analysis, we obtained imputed genotypes from 89% of the WTCCC case control subjects. Presumably, this reflects the influence of the low linkage disequilibrium that exists between rs3738919 and the genotyped SNPs (0.01 <r 2 Table 2) was performed using the actual genotypes from ITGAV SNPs rs3911238 and rs10174098 (Additional data file 1) and the imputed genotypes for rs3738919 (Table 1). One susceptibility haplotype (1-1-1), containing the major allele at each of the three SNPs, was significantly over-represented in cases compared with controls (OR = 1.15, 95% CI 1.02 to 1.30; P = 0.021).

Figure 1
Meta-analysis of the ITGAV single-nucleotide polymorphism rs3738919 in four independent rheumatoid arthritis Caucasian cohorts: Jacq and col-leagues [20], Wellcome Trust Case Control Consortium (WTCCC) [25], New Zealand and Oxford sample sets Meta-analysis of the ITGAV single-nucleotide polymorphism rs3738919 in four independent rheumatoid arthritis Caucasian cohorts: Jacq and colleagues [20], Wellcome Trust Case Control Consortium (WTCCC) [25], New Zealand and Oxford sample sets.  Variants rs3911238 and rs10174098 were typed over the NZ and Oxford sample sets, and association of three-marker haplotypes was examined (Additional data file 1 and Table 2). There was no support for a positive association of the 1-1-1 haplotype with RA in either the NZ or Oxford sample set. There was a significant protective effect for this haplotype in the Oxford samples (OR = 0.78, 95% CI 0.62 to 0.99; P = 0.042).

Meta-analysis of ITGAV in the four independent case control sample sets
Meta-analysis of all four sample sets, using a fixed effects model, revealed some evidence for an association of rs3738919 with RA (OR = 0.92, 95% CI 0.86 to 0.99; P = 0.021) (Figure 1a). Because of evidence for heterogeneity between the sample sets (Breslow-Day P = 0.011), a random effects model was also used for meta-analysis; this did not provide evidence for an association of rs3738919 with RA (OR = 0.92, 95% CI 0.80 to 1.07; P = 0.29) (Figure 1b). The Oxford sample set is significantly different from the other three sample sets at rs3738919; however, the Oxford patient sample set did not have any large differences in relation to gender (Oxford 28% male, NZ 34%, WTCCC 25%, France 13%), RF status (Oxford 77% positive, NZ 83%, WTCCC 84%, France 75%) or inheritance of the shared epitope (Oxford 77% positive, NZ 79%, WTCCC 79%, France 79%).

Stratification according to subphenotype
The NZ and Oxford samples were stratified according to gender, RF and shared epitope status ( Table 3). The stratification results did not reveal any significant differences in the rs3738919 genotype distribution for RA patients. There was a difference in the rs3738919 genotype distribution of male and female patients in the NZ cases (P = 0.026) (Table 3). However, a significant difference was not evident in either the Oxford or WTCCC case sample set (P = 0.55 and 0.80, respectively) ( Table 3).

Discussion
There was no evidence supporting a role for the ITGAV SNP rs3738919 in the etiology of RA when the NZ and the Oxford case control sample sets were analysed separately (P = 0.11 and 0.07, respectively) ( The rs3738919 ITGAV variant previously showed no association with RA in Japanese case-control samples [28]. These data were not included in our meta-analysis given that the major allele was present at a considerably different frequency in Japanese controls (0.92) than in Caucasian controls (France, 0.58; OXFORD, 0.64; and NZ, 0.63). It is already clear that there are genetic differences in susceptibility to RA between the Japanese and Caucasian populations. For example, the HLA-DRB1*0405 allele is most strongly associated with RA in Japanese patients [29] whereas the *0401 and *0404 alleles are more associated with RA in Caucasians [30]. Other genes showing population-specific effects in RA are the R620W variant of the protein tyrosine phosphatase, PTPN22, associated with Caucasian RA [4,5] but monomorphic in the Japanese population [31]; CTLA4, associated with RA in Caucasian [6] but not associated with RA in Japanese patients [32]; peptidylarginine deiminase type 4, PADI4, associated in Japanese patients [33,34] but very weakly in Caucasians [6, 35,36]; and the Fc receptor-like 3 gene variant, FCRL3-169C, associated with RA in Japanese patients [37,38] and in one Caucasian study [39] but not in other Caucasian studies [40,41]. In conclusion, we have not been able to provide further support for the involvement of ITGAV in the etiology of RA in Caucasians. However, it is important that further genotyping be done in a large independent cohort to confirm whether ITGAV plays a role in RA.

Conclusions
In Caucasians, meta-analysis of 3,527 cases and 4,126 controls does not provide further evidence for a role of the ITGAV SNP rs3738919 in the development of RA.