Figures
Abstract
Background
BANK1 and BLK belong to the pleiotropic autoimmune genes; recently, epistasis between BANK1 and BLK was detected in systemic lupus erythematosus. Although BLK has been reproducibly identified as a risk factor in rheumatoid arthritis (RA), reports are conflicting about the contribution of BANK1 to RA susceptibility. To ascertain the real impact of BANK1 on RA genetic susceptibility, we performed a large meta-analysis including our original data and tested for an epistatic interaction between BANK1 and BLK in RA susceptibility.
Patients and Methods
We investigated data for 1,915 RA patients and 1,915 ethnically matched healthy controls genotyped for BANK1 rs10516487 and rs3733197 and BLK rs13277113. The association of each SNP and RA was tested by logistic regression. Multivariate analysis was then used with an interaction term to test for an epistatic interaction between the SNPs in the 2 genes.
Results
None of the SNPs tested individually was significantly associated with RA in the genotyped samples. However, we detected an epistatic interaction between BANK1 rs3733197 and BLK rs13277113 (Pinteraction = 0.037). In individuals carrying the BLK rs13277113 GG genotype, presence of the BANK1 rs3733197 G allele increased the risk of RA (odds ratio 1.21 [95% confidence interval 1.04–1.41], P = 0.015. Combining our results with those of all other studies in a large trans-ethnic meta-analysis revealed an association of the BANK1 rs3733197 G allele and RA (1.11 [1.02–1.21], P = 0.012).
Citation: Génin E, Coustet B, Allanore Y, Ito I, Teruel M, Constantin A, et al. (2013) Epistatic Interaction between BANK1 and BLK in Rheumatoid Arthritis: Results from a Large Trans-Ethnic Meta-Analysis. PLoS ONE 8(4): e61044. https://doi.org/10.1371/journal.pone.0061044
Editor: Sunil K. Ahuja, South Texas Veterans Health Care System and University Health Science Center San Antonio, United States of America
Received: October 29, 2012; Accepted: March 5, 2013; Published: April 30, 2013
Copyright: © 2013 Génin et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was partially funded by the RETICS Program, RD08/0075 (RIER) from Instituto de Salud Carlos III, within the VI PN de I+D+i 2008–2011. Unrestricted grant from PFIZER, ROCHE, ABBOTT, UCB. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: Co-author Pr Philippe Dieudé received funding from a commercial source (PFIZER, ROCHE, ABBOTT, UCB), but this does not alter the authors' adherence to all PLOS ONE policies on sharing data and materials.
Introduction
Rheumatoid arthritis (RA) is a systemic, inflammatory, autoimmune disorder that affects up to 1% of the general adult population worldwide [1]. Nearly two-thirds of cases are seropositive for rheumatoid factor (RF) or anti-citrullinated protein autoantibodies (ACPA) [2]. RA is a complex disease implicating environmental, genetic and stochastic factors. However, the weight of genetics is underlined by family studies estimating that more than 50% of the variance in disease risk is explained by genetic factors [3]. To date, a genetic etiology for RA is unequivocal: recent genome-wide association studies have identified at least 31 validated susceptibility loci and implicated a broad array of biological pathways. These risk alleles explain about 16% of the disease variance, so additional susceptibility variants remain to be discovered [4].
Family-based epidemiology studies suggest a shared genetic basis among several autoimmune diseases leading to the paradigm of pleiotropy [5]. This genetic overlap is illustrated by the well-known association of certain human leukocyte antigen (HLA) loci and multiple human autoimmune diseases, as well as non-HLA risk loci in diverse pathways such as PTPN22, STAT4, IRAK1 or IRF5 (reviewed in [6]).
BANK1, encoding the B-cell scaffold protein with ankyrin repeats 1, is a Lyn tyrosine kinase substrate that promotes tyrosine phosphorylation of inositol 1,4,5-trisphosphate receptors [7]. BANK1 has been convincingly identified as a risk factor for the autoimmune diseases systemic lupus erythematosus (SLE) and systemic sclerosis (SSc) [8]–[12]. However, in terms of the contribution of BANK1 to RA susceptibility, case–control association studies of the European Caucasian population gave conflicting results [13]–[15]. The association signal for the human BANK1 gene is located between exons 1 and 3, with 2 polymorphisms, rs17266594 and rs10516487, in very high linkage disequilibrium (LD) with each other [8], [14]. Recently, the rs17266594 SNP, located in the branch-point consensus sequence, was found to have a negligible effect on splicing or gene expression, whereas the non-synonymous SNP rs10516487 (R61H) showed a dual nature by first influencing mRNA splicing and then protein quantity and second producing a risk-variant-containing protein isoform with increased potential for multimerization [16]. A third non-synonymous SNP, rs3733197, (A383T) located in exon 7 and encoding the ankyrin repeat-like motif, was found strongly associated with both SLE and SSc [8]–[11].
BLK was also recently identified as a susceptibility gene for multiple autoimmune phenotypic traits, including SLE [9], [17], [18], SSc [19]–[21], Sjögren's syndrome [22] and RA [15], [23]–[25]. BLK is a Src tyrosine kinase specifically expressed in the B-cell lineage [26]. The pleiotropic genetic variant of BLK, rs13277113, is located in the promoter of BLK, and the risk genotypes may be associated with reduced BLK mRNA levels in SLE [17].
Thus, both BANK1 and BLK, which encode for proteins involved in B-cell signaling, could be considered autoimmune pleiotropic genes. Of interest, a genetic and physical interaction of BANK1 and BLK was recently detected in SLE [27]. We aimed to re-evaluate the role of BANK1 in RA susceptibility and test whether BANK1– BLK epistatis could be involved in the RA genetic background, notably in the subset of patients positive for specific autoantibodies.
Materials and Methods
Patients and controls
A total of 1,454 RA patients (820 from France and 634 from Japan) and 1,542 ethnically matched healthy controls (1,220 from France and 322 from Japan) were genotyped for BANK1 rs10516487 and rs3733197 and BLK rs13277113. These original data were combined with already published data for 461 cases and 373 controls from Spain genotyped for the same 3 SNPs [13], for a total of 1,915 RA cases and 1,915 ethnically matched healthy controls. All RA patients were tested for 1) anti-CCP2 antibodies detected by commercial ELISA (Immunoscan, Eurodiagnostica, Arnheim, The Netherlands), and 2) RF by laser nephelometry.
All subjects provided informed, written consent as approved by the institutional review boards for the recruiting sites of the affiliated institutions (CPP IDF8, Paris, France; ethics committee, University of Granada, Spain; and ethics committee, University of Tsukuba, Japan).
To ascertain the real impact of BANK1 on RA susceptibility, we performed a meta-analysis including our original data and published data for 2,119 patients with RA and 2,163 controls from Spain, Sweden, Argentina, and Mexico [14]. For meta-analysis of BANK1 rs10516487, we included 5,325 RA patients and 11,780 controls from the United Kingdom: 1) 3,468 RA cases and 8,847 controls described in [15] with genotype and clinical information kindly provided by Dr. Orozco and 2) 1,857 RA cases and 2,933 controls from the Wellcome Trust Case Control Consortium [28]. All patients fulfilled the American College of Rheumatology 1987 revised classification criteria for RA [29].
Genotyping methods
Subjects were genotyped for BANK1 rs10516487 and rs3733197 and BLK rs13277113 by use of a competitive allele-specific PCR system (Kaspar genotyping, Kbioscience, Hoddeston, UK) and TaqMan 5′ discrimination assay (Applied Biosystem, Foster City, SA).
Statistical analysis
Statistical analyses involved use of R v2.13.1 (http://www.R-project.org, the R Foundation for Statistical Computing, Vienna, Austria). Tests of conformity to Hardy–Weinberg equilibrium involved a standard chi-square test (1 degree of freedom [df]) of differences between observed and expected genotype counts.
The association of SNPs and RA was tested by a logistic regression model assuming multiplicative effects (i.e., with genotype AA as reference, the risk associated with the homozygous GG genotype was assumed to be the square of the risk associated with the AG genotype). To test for interaction between the alleles for the two SNPs, we used a multivariate logistic regression model, with the 2 main effects for each of the SNPs (assuming multiplicative effects of the alleles at each locus) and an interaction parameter. Analysis was performed for each sample separately, the 2 European Caucasian samples combined, and the whole sample (after adjusting for country of origin). Significance of the interaction parameter was tested by a 1-df likelihood-ratio test. The corresponding P-values for epistasis were termed Pinteraction and the odds ratio of the interaction parameter ORinteraction. We performed a stratified analysis to estimate the ORs for BANK1 rs10516487 and rs3733197 alleles assuming a multiplicative model for subsets of subjects carrying the AA, AG or GG genotypes of BLK rs13277113.
Another test of interaction involved the Model Based Multifactor Dimensionality Reduction (MB-MDR) method [30] with R package MBMDR-3.0.2 [31]. This method is a dimension reduction method for exploring gene–gene interactions that involves merging multilocus genotypes into a 1-D construct that reflects the risk and can be represented by 3 values: high, low and non-informative. The method consists of 3 steps. In the first step, an association test is performed for each multilocus genotype and, depending on the resulting p value, the mutilocus genotype is assigned to 1 of the 3 categories: “high” if the p value is ≤0.1 and the score is positive, “low” if the p value is ≤0.1 and the score is negative and “undetermined” if the p value is >0.1. In a second step, multilocus genotypes of the same risk category are merged and 2 new association tests are performed, one for the high-risk category and one for the low-risk category. To account for multiple testing, the significance of the maximum of these two Wald statistics is then assessed by permutations.
We performed a meta-analysis of the BANK1 association with RA using the R package rmeta by pooling the results from the current study for BANK1 and results from the literature by searching the Medline database using the following keywords: rheumatoid arthritis, gene, BANK1 [32]. The DerSimonian-Laird random effects model [33] was used for allele counts in the different samples. Heterogeneity between samples was tested by the Q test [34] and the I2 statistic [35]. This study complies with the PRISMA statement [36].
Results
For both French and Japanese samples, the genotyping success rate was >95%, with no significant departure from Hardy–Weinberg equilibrium at p<0.05. In addition, control allelic frequencies for BANK1 rs10516487 and rs3733197 and BLK rs13277113 were similar to those reported for the European Caucasian and Japanese populations in the HapMap project (http://hapmap.ncbi.nlm.nih.gov/).
We had available genotype data for the 3 markers for 1,901 RA patients (806 from France, 461 from Spain and 634 from Japan) and 1,898 ethnically matched controls. The characteristics of RA patients are in Table 1.
Analysis of single BANK1 SNPs failed to detect an association with RA in each population investigated (i.e., French, Spanish and Japanese) or the combined population (Figure 1). Similarly, we found no association of BLK rs13277113 and RA in all populations investigated (Figure 1).
Number of cases and controls, frequency of G allele, odds ratio (OR) and 95% confidence interval (95% CI) and P-value of the association under a multiplicative model for each population separately, for European Caucasians and for the combined dataset (France, Spain and Japan).
We found an interaction between BANK1 rs3733197 and BLK rs13277113 in the European Caucasian population (ORinteraction = 1.23 [95% confidence interval (95% CI) 1.02–1.48], Pinteraction = 0.032). The risk of RA was increased with the G risk allele of BANK1 rs3733197 only for individuals carrying the GG homozygous genotype of BLK rs13277113 (ORG/GG = 1.22 [1.04–1.42], P = 0.013). We found no effect of the rs3733197 G allele in subjects carrying the rs13277113 AG or AA genotypes (ORG/AG = 1.02 [0.84–1.23], P = 0.86, and ORG/AA = 0.77 [0.49–1.21], P = 0.26, respectively; Table 2, Figure 2). BANK1 rs3733197 and BLK rs13277113 showed no epistasis in the Japanese population (ORinteraction = 1.15 [0.77–1.72], Pinteraction = 0.49). Meta-analysis of the combined sets (i.e., French, Spanish and Japanese populations) confirmed the epistatic interaction between BANK1 rs3733197 and BLK rs13277113 (ORinteraction = 1.17 [1.01–1.36], Pinteraction = 0.037), and results were homogeneous among the different populations (Q-test [2 df] = 0.85, P = 0.65, I2 = 0%). Concordant with results observed in the European Caucasian population, for the combined set, the association of the G risk allele of BANK1 rs3733197 and RA was restricted to subjects carrying the BLK rs13277113 GG genotype (ORG|GG(meta) = 1.21 [1.04–1.41], P = 0.015; Table 2, Figure 2).
OR for trend for the G allele of BANK1 rs3733197 and the three BLK rs13277113 genotypes by population (France, Spain and Japan).
The MB-MDR test for BANK1 rs3733197 and BLK rs13277113 also provided significant results for the European samples (France and Spain combined and adjusted on country of origin) leading to a Wald statistics maximum (Wmax) of 9.44 for the high-risk GG-GG genotype and a P-value of 0.036 (assessed with 500 permutations) with 95% CI 0.019–0.052. No findings were significant for the Japanese sample, and the GG-GG genotype combination was not classified as high risk (Table 3).
We found no significant interaction of BANK1 rs10516487 and BLK rs13277113, whatever the population investigated (data not shown).
The systematic review retrieved 2 studies [14], [15] (Figure S1). In combining our results with those of all association studies of the two BANK1 SNPs and RA in a large trans-ethnic meta-analysis, we found an association of the functional BANK1 rs3733197 polymorphism and RA, with the G allele conferring increased risk (OR = 1.11, [1.02–1.21], P = 0.012) (Figure 3). Similar ORs, although not significant, were obtained for non-European Caucasian populations (OR = 1.11 [0.86–1.43]; P = 0.42), probably because of smaller sample sizes than for the European Caucasian populations. The power (at p<0.05) to detect the effect of the G risk allele of BANK1 rs3733197 in the non-European populations was only 21.8% as compared with 60% for the European Caucasian populations. Effects were homogenous between all samples (Q-test = 5.97, P = 0.314; I2 = 16.3% [0%–78.8%]). Case–case analyses failed to detect any association with specific autoantibody status (ACPA and RF) (data not shown). Meta-analysis of the BANK1 rs10516487 variant (R61H), which is functional [15], remained inconclusive, with an estimated OR of 1.06 ([1.00–1.12], P = 0.055; I2 = 13% [0%–74.6%]) for the association of the G allele and RA risk. The results were similar when restricting the analysis to European Caucasian populations or non-European Caucasian populations.
Number of cases and controls; frequency of the G allele in cases and controls; and OR, 95% CI and P-value for the association for each population separately and for European Caucasians and non-European Caucasians and the combined dataset. All data were obtained from the literature, except for the France and Japan data, for which new genotyping was performed for this study.
Discussion
This study provides the first evidence of a contribution of the BANK1 gene to RA susceptibility. It suggests that BANK1 does not play a role in RA individually but only when interacting with BLK. In addition, this is the first report providing some evidence of epistasis between two RA susceptibility genes.
BLK has been convincingly established as an RA susceptibility gene, but the role of BANK1 was uncertain: results of a previous meta-analysis were inconclusive and prompted us to investigate ethnic groups not investigated in the first meta-analysis [13]. Although we failed to detect any significant individual association for the two BANK1 SNPs we investigated, the large trans-ethnic meta-analysis provided evidence for an association of BANK1 rs3733197 and RA (OR = 1.11, [1.02–1.21], P = 0.015) (Figure 3).
The clustering of multiple autoimmune disorders in families and evidence for autoimmune pleiotropic loci are well known. With evidence for an association of BANK1 and RA, both BANK1 and BLK should be considered pleiotropic genes involved in the genetic background of distinct autoimmune phenotypes including SLE, SSc, Sjögren's syndrome and now RA. Interestingly, a genetic interaction between BANK1 and BLK was recently identified in SLE [27]. Although the BANK1–BLK epistasis previously described in SLE involves different variants than those we investigated, our results strengthen the concept of pleiotropic genes and, even more importantly, pleiotropic genetic interactions pointing to a B-cell specific pathway that could be relevant in both SLE and RA pathogenesis.
BLK encodes a tyrosine kinase involved in the regulation of B-cell activation, and BANK1 encodes for a B-cell adaptor protein [7], [37]. B lymphocytes play several critical roles in the pathogenesis of RA. They are the source of RF and ACPA and are involved in antigen presentation to T cells and cytokine production. The success of B-cell depletion therapy in RA illustrates the importance of this specific lymphocyte compartment [38]. Despite lack of functional or mechanistic data, an important limitation of this study, our results revealed an epistasis between a functional BLK variant and a BANK1 non-synonymous variant (A383T) in exon 7, which encodes the ankyrin repeat-like motif [8], [17]. Consequently, both BANK1 and BLK disease-associated variants could contribute to sustained B-cell receptor signaling and B-cell hyperactivity. Further genetic association studies, involving large samples, are required to replicate our findings and better specify the putative role of BANK1–BLK epistasis in modulating the autoantibody status of RA, although we did not detect an association related to a particular ACPA status.
Of interest, Castillejo-López et al. recently demonstrated that B-cell receptor stimulation enhanced BANK1 and BLK binding [27], so genetics may guide the identification of true functional biological pathways in complex diseases [39]. Saijo et al. showed that mice triple deficient in the Src-family protein tyrosine kinases BLK and LYN but not single-deficient or SYK-deficient mice showed impaired NF-kappaB induction and B-cell development [40]. BANK1 contains several predicted motifs for binding the SH2 and SH3 domains of Src kinases. Therefore, BANK1–BLK epistasis may have specific functional consequences for B cells, which might be involved in RA pathogenesis.
In our samples of about 2,000 cases and 2,000 controls, we found no association of BANK1 or BLK and RA but rather a significant genetic interaction. Taking into account the growing evidence that epistasis contributes to risk for complex diseases [41]–[43], our findings illustrate that accounting for gene–gene interactions might be necessary to identify genetic effects that could otherwise be missed. Indeed, the univariate approach considering only a single marker at a time, commonly used in genome-wide association studies, could overlook the complex interactions that often occur in biological systems [44].
Unexpectedly, the BANK1–BLK epistasis we identified involves the BLK rs13277113 GG genotype found protective in SLE. Such an inverse association could be surprising, but there are several examples of variants having opposite effects in distinct auto-immune diseases [45]. Another striking feature of the data is the huge difference in BLK rs13277113 minor allele frequency between the European samples and the Japanese sample. The G allele is indeed the major allele in Europe and the minor allele in Japan. Because of this difference, which could be due to differential selection between populations of different ancestry, use of the Japanese population as a replication sample to assess the BANK1–BLK epistasis detected in the European samples is difficult. Indeed, in terms of the multilocus genotype distribution in the Japanese sample (Table 3), the frequency of the GG-GG genotype combination was similar in cases and controls. However, the OR for the G allele of BANK1 rs3733197 was increased for the GG genotype as compared with the AG and AA genotypes of BLK rs13277113. Consequently, independent replication is mandatory before definitely establishing the contribution of BANK1–BLK epistasis to RA susceptibility. Detecting and replicating true epistasis has been difficult and represents a challenge in deciphering the genetic architecture of complex diseases [41], [46].
In summary, BANK1 contributes to the RA genetic susceptibility background, but the association may depend on the BLK rs13277113 genotype, in that our results suggest an epistasis between BANK1 and BLK. This epistasis could explain the discrepancies in the various reported association studies and suggest that BANK1 rs3733197 status should be re-analyzed for BLK rs13277113 status. Our findings illustrate the concept of pleiotropic epistatic interaction and suggest that BANK1 and BLK might play a role in RA pathogenesis.
Supporting Information
Figure S1.
Prisma Flow diagram summarizing how the meta-analysis was conducted.
https://doi.org/10.1371/journal.pone.0061044.s001
(DOCX)
Acknowledgments
The authors thank Dr Joëlle Benessiano (Centre de Ressource Biologique, Hôpital Bichat Claude-Bernard, Paris France) and staff members of the Établissement Français du Sang for their assistance, and Mrs Corine Bensimon for expert secretarial assistance.
We also thank Prof. Bernard Grandchamp and Pr. Nadem Soufir (Hôpital Bichat Claude-Bernard) for assistance in setting up the French Caucasian control sample.
The authors are indebted to Professor Anne Barton, Professor Jane Worthington and Doctor Gisela Orozco for providing BANK1 rs10516487 genotyping data performed in the UK population; to Professor Takayuki Sumida (Faculty of Medicine, University of Tsukuba) for recruiting study participants; and to Dr. Shomi Oka (Clinical Research Center for Allergy and Rheumatology, Sagamihara Hospital, National Hospital Organization) for technical assistance.
Author Contributions
Conceived and designed the experiments: EG PD NT JM. Performed the experiments: EG PD NT JM MT BC II. Analyzed the data: EG PD NT JM MT BC. Contributed reagents/materials/analysis tools: YA II AC TS ARW ST AC OV TB XLL PF HF OM BFG AB MAGG GC. Wrote the paper: EG PD NT JM.
References
- 1. Silman AJ, Pearson JE (2002) Epidemiology and genetics of rheumatoid arthritis. Arthritis Res 4 Suppl 3: S265–272.
- 2. Klareskog L, Catrina AI, Paget S (2009) Rheumatoid arthritis. Lancet 373: 659–672.
- 3. MacGregor AJ, Snieder H, Rigby AS, Koskenvuo M, Kaprio J, et al. (2000) Characterizing the quantitative genetic contribution to rheumatoid arthritis using data from twins. Arthritis Rheum 43: 30–37.
- 4. Stahl EA, Raychaudhuri S, Remmers EF, Xie G, Eyre S, et al. (2010) Genome-wide association study meta-analysis identifies seven new rheumatoid arthritis risk loci. Nat Genet 42: 508–514.
- 5. Criswell LA, Pfeiffer KA, Lum RF, Gonzales B, Novitzke J, et al. (2005) Analysis of families in the multiple autoimmune disease genetics consortium (MADGC) collection: the PTPN22 620W allele associates with multiple autoimmune phenotypes. Am J Hum Genet 76: 561–571.
- 6. Scott IC, Steer S, Lewis CM, Cope AP (2011) Precipitating and perpetuating factors of rheumatoid arthritis immunopathology: linking the triad of genetic predisposition, environmental risk factors and autoimmunity to disease pathogenesis. Best Pract Res Clin Rheumatol 25: 447–468.
- 7. Yokoyama K, Su Ih IH, Tezuka T, Yasuda T, Mikoshiba K, et al. (2002) BANK regulates BCR-induced calcium mobilization by promoting tyrosine phosphorylation of IP(3) receptor. EMBO J 21: 83–92.
- 8. Kozyrev SV, Abelson AK, Wojcik J, Zaghlool A, Linga Reddy MV, et al. (2008) Functional variants in the B-cell gene BANK1 are associated with systemic lupus erythematosus. Nat Genet 40: 211–216.
- 9. Guo L, Deshmukh H, Lu R, Vidal GS, Kelly JA, et al. (2009) Replication of the BANK1 genetic association with systemic lupus erythematosus in a European-derived population. Genes Immun 10: 531–538.
- 10. Chang YK, Yang W, Zhao M, Mok CC, Chan TM, et al. (2009) Association of BANK1 and TNFSF4 with systemic lupus erythematosus in Hong Kong Chinese. Genes Immun 10: 414–420.
- 11. Dieude P, Wipff J, Guedj M, Ruiz B, Melchers I, et al. (2009) BANK1 is a genetic risk factor for diffuse cutaneous systemic sclerosis and has additive effects with IRF5 and STAT4. Arthritis Rheum 60: 3447–3454.
- 12. Rueda B, Gourh P, Broen J, Agarwal SK, Simeon C, et al. (2010) BANK1 functional variants are associated with susceptibility to diffuse systemic sclerosis in Caucasians. Ann Rheum Dis 69: 700–705.
- 13. Suarez-Gestal M, Calaza M, Dieguez-Gonzalez R, Perez-Pampin E, Pablos JL, et al. (2009) Rheumatoid arthritis does not share most of the newly identified systemic lupus erythematosus genetic factors. Arthritis Rheum 60: 2558–2564.
- 14. Orozco G, Abelson AK, Gonzalez-Gay MA, Balsa A, Pascual-Salcedo D, et al. (2009) Study of functional variants of the BANK1 gene in rheumatoid arthritis. Arthritis Rheum 60: 372–379.
- 15. Orozco G, Eyre S, Hinks A, Bowes J, Morgan AW, et al. (2011) Study of the common genetic background for rheumatoid arthritis and systemic lupus erythematosus. Ann Rheum Dis 70: 463–468.
- 16. Kozyrev SV, Bernal-Quiros M, Alarcon-Riquelme ME, Castillejo-Lopez C (2011) The dual effect of the lupus-associated polymorphism rs10516487 on BANK1 gene expression and protein localization. Genes Immun
- 17. Hom G, Graham RR, Modrek B, Taylor KE, Ortmann W, et al. (2008) Association of systemic lupus erythematosus with C8orf13-BLK and ITGAM-ITGAX. N Engl J Med 358: 900–909.
- 18. Yang W, Ng P, Zhao M, Hirankarn N, Lau CS, et al. (2009) Population differences in SLE susceptibility genes: STAT4 and BLK, but not PXK, are associated with systemic lupus erythematosus in Hong Kong Chinese. Genes Immun 10: 219–226.
- 19. Ito I, Kawaguchi Y, Kawasaki A, Hasegawa M, Ohashi J, et al. (2010) Association of the FAM167A-BLK region with systemic sclerosis. Arthritis Rheum 62: 890–895.
- 20. Gourh P, Agarwal SK, Martin E, Divecha D, Rueda B, et al. (2010) Association of the C8orf13-BLK region with systemic sclerosis in North-American and European populations. J Autoimmun 34: 155–162.
- 21. Coustet B, Dieude P, Guedj M, Bouaziz M, Avouac J, et al. (2011) C8orf13-BLK is a genetic risk locus for systemic sclerosis and has additive effects with BANK1: results from a large french cohort and meta-analysis. Arthritis Rheum 63: 2091–2096.
- 22. Nordmark G, Kristjansdottir G, Theander E, Appel S, Eriksson P, et al. (2011) Association of EBF1, FAM167A(C8orf13)-BLK and TNFSF4 gene variants with primary Sjogren's syndrome. Genes Immun 12: 100–109.
- 23. Gregersen PK, Amos CI, Lee AT, Lu Y, Remmers EF, et al. (2009) REL, encoding a member of the NF-kappaB family of transcription factors, is a newly defined risk locus for rheumatoid arthritis. Nat Genet 41: 820–823.
- 24. Ito I, Kawasaki A, Ito S, Kondo Y, Sugihara M, et al. (2010) Replication of association between FAM167A(C8orf13)-BLK region and rheumatoid arthritis in a Japanese population. Ann Rheum Dis 69: 936–937.
- 25. Deshmukh HA, Maiti AK, Kim-Howard XR, Rojas-Villarraga A, Guthridge JM, et al. (2011) Evaluation of 19 autoimmune disease-associated loci with rheumatoid arthritis in a colombian population: evidence for replication and gene-gene interaction. J Rheumatol 38: 1866–1870.
- 26. Dymecki SM, Niederhuber JE, Desiderio SV (1990) Specific expression of a tyrosine kinase gene, blk, in B lymphoid cells. Science 247: 332–336.
- 27. Castillejo-Lopez C, Delgado-Vega AM, Wojcik J, Kozyrev SV, Thavathiru E, et al. (2012) Genetic and physical interaction of the B-cell systemic lupus erythematosus-associated genes BANK1 and BLK. Ann Rheum Dis 71: 136–142.
- 28. Consortium TWTCC (2007) Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 447: 661–678.
- 29. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, et al. (1988) The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 31: 315–324.
- 30. Calle ML, Urrea V, Vellalta G, Malats N, Steen KV (2008) Improving strategies for detecting genetic patterns of disease susceptibility in association studies. Stat Med 27: 6532–6546.
- 31. Calle ML, Urrea V, Malats N, Van Steen K (2010) mbmdr: an R package for exploring gene-gene interactions associated with binary or quantitative traits. Bioinformatics 26: 2198–2199.
- 32.
Lumley T (2009) rmeta: Meta-analysis: R package website. Available: http://CRAN.R-project.org/package=rmeta. Accessed 2013 March 22.
- 33. DerSimonian R, Laird N (1986) Meta-analysis in clinical trials. Control Clin Trials 7: 177–188.
- 34. Cochran WG (1954) The combination of estimates from different experiments. Biometrics 10: 101–129.
- 35. Higgins JP, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21: 1539–1558.
- 36. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, et al. (2009) The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med 6: e1000100.
- 37. Kurosaki T (2002) Regulation of B-cell signal transduction by adaptor proteins. Nat Rev Immunol 2: 354–363.
- 38. Edwards JC, Cambridge G (2006) B-cell targeting in rheumatoid arthritis and other autoimmune diseases. Nat Rev Immunol 6: 394–403.
- 39. Sun YV, Kardia SL (2010) Identification of epistatic effects using a protein-protein interaction database. Hum Mol Genet 19: 4345–4352.
- 40. Saijo K, Schmedt C, Su IH, Karasuyama H, Lowell CA, et al. (2003) Essential role of Src-family protein tyrosine kinases in NF-kappaB activation during B cell development. Nat Immunol 4: 274–279.
- 41. Cordell HJ (2009) Detecting gene-gene interactions that underlie human diseases. Nat Rev Genet 10: 392–404.
- 42. Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, et al. (1993) Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science 261: 921–923.
- 43. Evans DM, Marchini J, Morris AP, Cardon LR (2006) Two-stage two-locus models in genome-wide association. PLoS Genet 2: e157.
- 44. Moore JH (2003) The ubiquitous nature of epistasis in determining susceptibility to common human diseases. Hum Hered 56: 73–82.
- 45. Sirota M, Schaub MA, Batzoglou S, Robinson WH, Butte AJ (2009) Autoimmune disease classification by inverse association with SNP alleles. PLoS Genet 5: e1000792.
- 46. Reimherr M, Nicolae DL (2011) You've gotta be lucky: Coverage and the elusive gene-gene interaction. Ann Hum Genet 75: 105–111.