Susceptibility of autoimmune diseases in three polymorphisms of infection-associated gene IRAK 1

1 Nanfang Hospital, Southern Medical University, Guangzhou, China 2 School of Traditional Chinese Medicine, Southern Medical University, Guangzhou, China 3 The First College of Clinical Medicine, Shandong University of Traditional Chinese Medicine, Jinan, China 4 Department of Medicine, Virginia Commonwealth University and Massey Cancer Center, Richmond, Virginia, USA 5 Department of Intensive Care Unit, Chinese PLA General Hospital, Beijing, China 6 School of Biotechnology, Southern Medical University, Guangzhou, China


Introduction
Autoimmune diseases (ADs) are a group of complex disorders initiated by the loss of tolerance to self-antigen, which results in immune-mediated tissue destruction and chronic disabilities [1].ADs comprise more than 100 diseases and syndromes, and the annual estimated treatment costs for ADs are more than US$100 billion [2].As a group of complex diseases, the precise molecular mechanism of ADs is still not clear.However, the interaction of genes and environment is widely recognized as one of the main causes of ADs [3,4].
The sustained pathology of ADs is directly caused by a specific self-reactive immune response, including innate and adaptive immune response, which can be caused by infection with some kinds of pathogenic microorganisms [5][6][7].Interleukin-1 (IL-1) receptorassociated kinases (IRAKs) are key mediators in the signaling pathways of innate immune response, especially in the Toll-like receptors (TLRs)/IL-1 receptors (IL-1Rs) pathway.There are four kinds of IRAKs: IRAK1, IRAK2, IRAK3, and IRAK4 [8].IRAK1 is the first member identified in the IRAK family.It can be phosphorylated and induce a serious downstream signaling cascade after the activation of TLRs/IL-1Rs stimulation [8,9].The phosphorylation of IRAK1 is associated with the activation of NF-κB in inflammatory disease, and the activity of NF-κB can be inhibited using an IRAK1 inhibitor, resulting in the suppression of the inflammatory conditions [10,11].IRAK1 has been found play an important role in both ADs patients and in an autoimmune animal model [11][12][13][14][15]. Therefore, IRAK1 is recognized as a risk gene in ADs.
Single nucleotide polymorphisms (SNPs), or mutations, may alter expression of the gene and influence the susceptibility of some diseases [16][17][18][19].Some researchers have studied the relationship between ADs risk and three polymorphisms of IRAK1: IRAK1 rs3027898 C>A, IRAK1 rs1059702 T>C, and IRAK1 rs1059703 T>C.Most of these studies were conducted in developing countries, so it is very important for these countries to make clear what the role of IRAK1 for ADs is [14,15,[20][21][22][23][24][25][26][27].However, the results among these studies remain conflicting.Therefore, we conducted this study, according the procedure published by MOOSE group [28], to find a clearer association between these three SNPs and ADs risk.

Publication search
A systematic search was performed in PubMed, OvidSP, and Chinese National Knowledge Infrastructure (CNKI) databases covering all papers published prior to August 2014.The searching strategy was as follows: (autoimmune OR autoimmune disease OR autoimmunity) AND (polymorphism OR polymorphisms OR variation OR variations OR mutation OR mutations OR variant OR variants) AND (IRAK1 OR rs3027898 OR rs1059702 OR rs1059703).The references in the studies were also read to find additional publications on this topic.Articles included had to meet the following criteria: case-control study; evaluation of IRAK1 polymorphisms (rs3027898, rs1059702, or rs1059703) and risk of ADs; and available and usable data of genotype frequency.

Data extraction
Two authors independently extracted the data from eligible studies.The different data that were extracted were checked.The remaining disagreements were discussed and judged.The following information was extracted: first author, publication year, diseases, country, ethnicity, genotyping methods, number of cases and controls, gender distribution of cases and controls, number of genotypes and alleles, Hardy-Weinberg equilibrium (HWE) in control subjects, and frequency of major allele in controls.Ethnicities were categorized as Caucasian, Asian, African, and Latin-American.Study quality was judged according to the criteria modified from previous publications [29][30][31] (See Supplementary "Table S1 Scale for methodological quality assessment").

Statistical analysis
Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated as a measure of the association between the three SNPs (rs3027898, rs1059702, and rs1059703) and ADs risk.An allele model and other types of genetic models (heterozygote, homozygote, dominant, and recessive models) were used.In addition to comparing among all subjects, the stratified comparisons were also used according to different ethnicities and different diseases.The between-study heterogeneity was measured by Cochran's (Q) and Higgins's (I 2 ) tests.If the heterogeneity was considered significant (p < 0.05), the random effects model was used to estimate the pooled OR.Otherwise, the fixed effects model was used.Also, logistic metaregression analysis was carried out, if there was obvious significant heterogeneity, to explore potential sources of heterogeneity.The examined characteristics included publication years, countries, genotyping methods, number of alleles and genotypes, number of females and males in cases, and the frequency of major allele in SNP in controls.The HWE was examined using the Chi-square test with significance set at p < 0.05.Sensitivity analysis was performed to evaluate the effect of each study on the combined ORs by deleting each study in each turn.Potential publication bias was determined using Funnel plots and Begg's test.An asymmetric plot and p value of less than 0.05 was recognized as significance.All statistical analyses were performed using STATA 12.0 software.

Study characteristics
There were 483 articles matching the search strategy, and an additional article [20] was found by scanning the references of the original papers.After a step-by-step screening of the titles, abstracts, and full texts of the articles, as shown in Figure 1, there were 10 articles appropriate for this meta-analysis, which included 11 studies of rs3027898, 9 studies of rs1059702, and 7 studies of rs1059703.
Within all the 10 articles, six kinds of genotyping methods were used.Four races were included: Caucasian, Asian, African, and Latin-American.Four studies were not in HWE in control groups [14,15,23,24].In 484 articles, 34 were found not related to ADs and 145 were found not related to IRAK1 by scanning the titles.After that, 245 articles were recognized as reviews, 32 were found not related to human patients and 14 articles were repeated papers by reviewing the abstracts.The full-text of the left 14 articles were carefully reviewed, in which 1 article was found not include usable data and 3 articles were found not about rs3027898, rs1059702 or rs1059703.At last, 10 articles were remained for this metaanalysis, which included 11 case-control studies for rs3027898, 9 studies for rs1059702 and 7 studies for rs1059703.There was not enough data in another article [25] to generate the HWE in four studies, but the p value of HWE was not less than 0.001 according to the authors' explanation.The detail characteristics are shown in Table 1.

Association between IRAK1 rs1059703 T>C polymorphism and ADs risk
There was no significant increased risk in overall comparison in any genetic model of association between rs1059703 T>C polymorphism and the risk of ADs.However, the increased risk could be found in subgroup analysis based on ethnicities or diseases.In Caucasians, there was a significant increased risk of C allele with ADs (C versus T: OR = 1.35, 95% CI = 1.24-1.47,p = 0.000) (Table 4 and Figure S3A).In Asians, there was significant increased risk of TT allele with ADs in the heterozygote model (TC versus TT: OR = 0.77, 95% CI = 0.64-0.92,p = 0.005) and dominant model (TC+CC versus TT, OR = 0.77, 95% CI = 0.64-0.91,p = 0.003) (Table 4 and Figure S3C, S3D).In the SLE subgroup, there was an increased disease risk in C allele in the allele model (C versus T: OR = 1.47, 95% CI = 1.33-1.61,p = 0.000) (Table 4 and Figure S3E).

Evaluation of heterogeneity
The heterogeneities among studies were obvious in the overall comparisons (rs3027898: I 2 = 83.6%,Tau 2 = 0.033, p = 0.000; rs1059702: I 2 = 81.4%,Tau 2 = 0.020, p = 0.000; rs1059703: I 2 = 89.6%,Tau 2 = 0.059, p = 0.000).The meta-regression analyses were conducted to further explore sources of heterogeneity.Several factors were tested as potential sources of heterogeneity, including publication years, countries, genotyping methods, number of genotypes and alleles, number of females and males in cases, and the frequencies of major alleles for each SNP in controls.As a result, the heterogeneities could not be explained by any of the potential sources by meta-regression analysis.

Sensitivity and publication bias analysis
The sensitivity analysis to test the influence of a single study on the overall meta-analysis was performed by deleting each study one at a time.As a result, the pooled estimate did not show significant difference, which indicated that the results were reliable statistically.No evidence of publication bias was found in current meta-analysis, identified by the Funnel plots, Egger's test (p = 0.986 for rs3027898; p = 0.875 for rs1059702; p = 0.596 for rs1059703), and Begg's test (p = 0.533 for rs3027898; p = 0.917 for rs1059702; p = 0.230 for rs1059703) (Figure 3).

Discussion
IRAK1 is a protein kinase involved in the Toll/IL-1 receptor (TIR) pathway [32], which plays an important role in the activation of NF-κB.By enhancing the communication of TLR with TNF receptor-associated factor (TRAF) 6, or by engaging into the MyD88-signaling complex, IRAK1 could trigger NF-κB, subsequently increasing the expression level of several inflammatory cytokines, such as TNFα and IL-8 [33][34][35].Several animal experiments showed that the expression level of IL-17 was decreased and inflammatory responses were dampened by depletion of IRAK1 [36], and IRAK1 -/-mice were protected from experimental autoimmune encephalomyelitis (EAE) [13].Moreover, IRAK1 was found to be correlated with ADs risk in several studies of patient cohorts [12,20,22].
Three SNPs of IRAK1 have been found to be related to ADs risk: rs3027898 for RA [15,22], rs1059702 for SSc [23,24] and rs1059703 for SLE [14,25].However, the results remain in conflict.Therefore, we conducted this meta-analysis to better understand whether these three SNPs contribute to susceptibility to ADs.
In this meta-analysis, we screened 10 manuscripts and pooled the corresponding data, including 10,705 cases (9,865 controls) for rs3027898, 15,005 cases (14,997 controls) for rs1059702, and 8,115 cases (6,815 controls) for rs1059703.We found that all these three SNPs were related to ADs risk.
For rs3027898, C allele or CC genotype were correlated with increased disease risk in most of the genetic models, including the allele model, heterozygote model, and dominant model, both in pooled comparison and in the Caucasian subgroup.Moreover, the increased disease risk of CC genotype was also found in the homozygote model in the Caucasian subgroup.However, in the Asian subgroup, the increased ADs risk of CC genotype could only be found in the heterozygote model.In the stratified analyses based on different types of ADs, the increased susceptibility of CC genotype was found in the heterozygote model in the RA subgroup.Due to the data limitation of the SLE subgroup, we could only compare the association in the allele model, and indeed found the increased risk of C allele.
For rs1059702, either in pooled or in stratified analyses, the increased disease risk for T allele or TT genotype was found in the allele, homozygote, and dominant models.In the heterozygote model, the increased disease risk of TT genotype was found both in pooled analysis and in the Asian subgroup.In the recessive model, compared with CC genotype, the TT+TC genotype was found to be associated with increased disease risk in the Asian subgroup and SSc subgroup.
For rs1059703, there was not as much association as with rs3027898 or rs1059702.No significant relationship between ADs risk and rs1059703 could be found in pooled analyses in any genetic model.However, there were some associations shown when the stratified analyses were done.In the allele model, increased disease risk with C allele was found in the Caucasian and SLE subgroups.In contrast, increased ADs risk was found to be associated with TT genotype in the Asian subgroup, both in the heterozygote and dominant models.But for the RA subgroup, no association was found in any genetic model.
There are some limitations in this study.First, although 10 articles were included, the studies for some stratified analyses were limited.For example, there were only two studies of the Asian subgroup and two studies of the SSc subgroup in analyses for rs1059702, except in the allele model.Second, there is obvious heterogeneity between different groups for some genetic models.Although the meta-regression and sensitivity analyses were conducted and no potential source of heterogeneity was found, the results still must be treated with caution.Third, only three SNPs in IRAK1 were included in this study.However, there are more SNPs in IRAK1 and more genes in the TIR signaling pathway, which would also contribute to susceptibility of ADs.The effect of these SNPs and genes, and also the interaction or network among these genetic locations, should be studied in the future.Furthermore, studies investigating the geneenvironment interactions will also help to make clear of the role of these SNPs in the pathogen of ADs [37][38][39][40].Finally, since ADs comprised diverse diseases, the relationship of these SNPs with other types of ADs, such as inflammatory bowel disease and seronegative spondyloarthropathies, should be investigated in the future.

Conclusions
The present study demonstrated that three SNPs (rs3027898, rs1059702 and rs1059703) in IRAK1 confer risk of ADs.Moreover, the associations were only within a specific genetic model, specific ethnicities, or specific disease types, not within all types of cohorts or ADs.

Figure 1 .
Figure 1.Flowchart for identification of studies included in the meta-analysis

Figure 2 .
Figure 2. Forest plots of overall analysis of ADs risk associated with IRAK1

Figure 3 .
Figure 3. Publication bias on the IRAK1 polymorphism and ADs risk

Table 1 .
Characteristics of published studies of rs3027898, rs1059702

Table 2 .
Stratified analysis of association between ADs risk and rs3027898

Table 3 .
Stratified analysis of association between ADs risk and rs1059702

Table 4 .
Stratified analysis of association between ADs risk and rs1059703