Introduction

Rheumatoid arthritis (RA) is a chronic systemic autoimmune disease characterized by multiple joints synovitis leading to joint destruction, disability and a poor quality of life. Both genetic and environmental factors play important roles in the pathogenesis of RA [1]. Large genome-wide association studies (GWAS) have identified more than 30 loci involved in RA pathogenesis [2]. Genetic factors including human leukocyte antigen gene (HLA) alleles such as HLA-DRB1*01, HLA-DRB1*13, HLA-DRB1*15, etc. [3], and non-HLA genes such as peptidylarginine deiminase type 4 gene (PADI4) [4], protein tyrosine phosphatase non-receptor 22 gene (PTPN22) [5], interleukin-23 receptor gene (IL-23R) [6], interleukin-1β gene (IL-1B) [7], macrophage migration inhibitory factor gene (MIF) [8], tumor necrosis factor alpha gene (TNF-α) [9] and CD40 [10] have been implicated in the pathogenesis of RA.

In RA, IL-18 is significantly elevated in sera, synovial fluid,and synovial tissues compared with those from osteoarthritic patients and healthy people [11], which is produced by synovial macrophages, synovial fibroblasts, endothelial cells, dendritic cells, articular chondrocytes, osteoblasts, and synovial fluid neutrophils [12]. IL-18 and other cytokines such as TNF-α, vascular endothelial growth factor (VEGF), IL-10 and IL-1 can activate nuclear factor-κB (NF-κB), promote expression of cyclooxygenase-2 (COX-2) and inducible nitric oxide synthase (iNOS), and induce synthesis of prostaglandin E2 (PGE2) and nitric oxide (NO), which lead to synovial inflammation associated with arthrosis, swelling, hyperplasia, angiogenesis, and finally cartilage degeneration [13–16].

IL-18, known as IFN-gamma-inducing factor, is a cytokine of the IL-1 family. The human IL-18 gene is located on chromosome 11q22.2-q22.3, and is composed of six exons and five introns [12]. Many polymorphisms in the promoter region of the IL-18 gene have been identified. Recently, these polymorphisms have attracted widespread attention, especially the IL-18 gene promoter −607A/C and −137C/G polymorphisms. IL-18 polymorphisms in the promoter region has been reported to be associated with many kinds of diseases, such as asthma [17], type 1 diabetes [18], breast cancer [19], adult-onset Still’s disease [20], and virus infective disease [21]. Furthermore, some studies suggested that IL-18 gene promoter polymorphisms are associated with RA, but some studies failed to find any association. In this paper, to derive a more precise estimation of the relationship between IL-18 gene promoter −607A/C, −137C/G polymorphisms and RA, a meta-analysis was performed.

Materials and methods

Search strategy

We conducted a computer-based searches of PubMed, EMBASE, Spring-link, Web of Science, China National Knowledge Infrastructure (CNKI) and Wanfang (Chinese) to identify all studies examining the association of IL-18 promoter −607A/C (rs1946518) and −137C/G (rs187238) polymorphisms with RA susceptibility with the last report up to January 2013. The following keywords and subject terms were used for searching: “rheumatoid arthritis”, “RA”, “Interleukin-18”, “IL-18”, “genetic variant”, “genetic variation” and “polymorphism”. No language restrictions were applied.

Inclusion and exclusion criteria

The following inclusion criteria were used to select literatures for the meta-analysis: (1) investigation of the IL-18 promoter gene −607A/C and −137C/G polymorphisms and RA susceptibility, (2) only the case–control studies were considered, (3) the papers should clearly describe RA diagnoses and the sources of cases and controls, (3) the authors must offer the size of the sample, OR and their 95 % CI or the information that could help to infer the results in the papers. When a study reported the results on different subpopulation, we treated them as a separated study in the meta-analysis. The exclusion criteria were: (1) none-case–control studies, (2) studies that contained overlapping data, (3) review articles.

Data extraction

Two investigators reviewed and extracted information from all eligible publications independently, according to the inclusion and exclusion criteria listed above. An agreement was reached by discussion between the two reviewers whenever there was a conflict. The following items were collected from each study: first author’s surname, year of publication, study population (country), ethnicity, numbers of cases and controls for the studies, genotypic methods. The ethnicity was classified as Asian or Caucasian.

Statistical analysis

The Hard-Weinberg Equilibrium (HWE) was measured by Chi squared test for the control groups of each study. Studies with control groups that were not in HWE (P < 0.05) were excluded.

The strength of associations between IL-18 promoter −607A/C and −137C/G polymorphisms and RA risk were measured by ORs with 95 %CIs. Meta-analyses was performed using (1) allele model (−607A/C: A vs. C, and −137C/G: C vs. G), (2) codominant model (−607A/C: AA vs. CC, and −137C/G: CC vs. GG), (3) dominant model (−607A/C: AA + AC vs. CC, and −137C/G: CC + CG vs. GG), and (4) recessive model (−607A/C: AA vs. AC + CC, and −137C/G: CC vs. CG + GG). Heterogeneity among studies was assessed by the Chi squared-based Q-statistic test and I 2 Statistics, if P < 0.10 and I 2 > 50 %, was considered significant. When the effects were assumed to be homogenous, fixed-effects model was used (the Mantel–Haenszel method) [22]; otherwise, random-effects model (DerSimonian and Laird method) [23] was conducted. Meta-regression was performed with ethnicity, publication year, genotypic method and sample size to explore reasons for heterogeneity among the studies. Sensitivity analysis was performed to evaluate the influence of single studies on the overall estimate. Begg’s funnel plots [24] and Egger’s test [25] were used to assesse publication bias, and P < 0.05 was indicated significant.

All statistical tests performed in the study were two-tailed. All analyses were done using STATA 12.0 (STATA Corporation, College Station, TX, USA).

Results

Study characteristics

The combined search yielded 48 references. Study selection process was shown in Fig. 1. Two studies (Sivalingam’s [26] study which conducted in Chinese population and Rueda’s [27] study) with control groups of −607A/C that were not in HWE, so they were excluded. Finally, a total of 12 studies in nine articles were included with 3,041 patients and 2,646 controls in our meta-analysis [26–34]. For the −607A/C, a total of 10 studies from eight articles involved 2,662 cases and 2,168 controls. For the −137C/G, a total of 9 studies from six articles involved 1,331 cases and 1,468 controls. In our meta-analysis, 8 studies were conducted in Asian populations and 4 studies conducted in Caucasian populations. Of the articles, 6 were published in English and 4 in Chinese. In all studies, three genotypic methods were used, including Sequence-specific PCR (PCR-SSP), polymerase chain reaction-restriction fragment length polymorphism (PCR–RFLP) and Taqman. In Gracie’s [28] study, Taqman was used to detect −607A/C polymorphism and PCR-SSP was used to detect −137C/G polymorphism. The characteristics of those studies in the meta-analysis were listed in Table 1.

Fig. 1
figure 1

Process of study selection

Full size image
Table 1 Characteristics of eligible studies
Full size table

Meta-analysis results and heterogeneity analysis

The results of this meta-analysis were shown in Table 2. For IL-18 −607A/C polymorphism with RA risk, significant association was found in allele model (OR = 0.778, 95 % CI = 0.633–0.955) and dominant model (OR = 0.618, 95 % CI = 0.466–0.819). However, no significant association could be observed between −137C/G and RA risk under all genetic models (allele model: OR = 0.940, 95 % CI = 0.777–1.138; codominant model: OR = 1.079, 95 % CI = 0.574–2.029; dominant model: OR = 0.913, 95 % CI = 0.779–1.069; recessive model: OR = 1.133, 95 % CI = 0.586–2.190). In the subgroup analysis based on ethnicity, significant result was also found in Asian population under allele model (OR = 0.717, 95 % CI = 0.527–0.977) and dominant model (OR = 0.541, 95 % CI = 0.417–0.702), but not found in Caucasian population for the relationship of IL-18 −607A/C polymorphism with RA risk; while no obvious association was found in Asian populations or Caucasian populations between IL-18 −137C/G polymorphism with RA risk.

Table 2 Meta-analysis of the IL-18 gene polymorphisms and RA risk
Full size table

There was significant heterogeneity in most comparison models. To explore sources of heterogeneity, we did meta-regression analyse of ethnicity, publication year, genotypic method and sample size, but we failed to identify the sources of heterogeneity.

Sensitivity analysis

Sensitivity analyses were conducted to assess the influence of each individual study on the pooled OR by deleting any single study each time. When Sugiura’s [31] study was excluded, the pooled result was significant for AA versus CC model of −607A/C in Asian subgroup (OR = 0.406, 95 % CI = 0.172–0.958). In the dominant model for −607A/C, when two of Sivalingam’s [26] studies and Sugiura’s [31] study were included, the pooled result was also significant in total group (OR = 0.667, 95 % CI = 0.483–0.920) and in Asian subgroup (OR = 0.628, 95 % CI = 0.396–0.995). The result of this meta-analysis is not stable just in codominant model for −607A/C polymorphism, it doesn’t affect the conclusion.

Evaluation of publication bias

Begg’s funnel plot and Egger’s test were performed to assess publication bias of the studies. We found no asymmetry of the funnel plot, and Egger’s test suggested that no publication bias was detected in any comparison model (P > 0.05) except the dominant model for −607A/C when all 10 studies were included. But two of Sivalingam’s [26] studies and Sugiura’s [31] study were excluded, no publication bias was observed. Data were shown in Table 2.

Discussion

The imbalance of Th1/Th2 is suggested to be one important mechanism in the induction and development of RA. IL-18 promotes upregulation of Th1 cells and inhibits Th2 cells in RA patient resulting in a greater imbalance of Th1/Th2 and aggravating patient illnesses [11]. IL-18 protein expression is regulated by the IL-18 promoter gene [35]. Three single nucleotide polymorphisms (SNPs) were detected at positions −137C/G, −607A/C and −656G/T within the IL-18 promoter region. Two of these, −137C/G and −607A/C have been extensively studied, which are located at the binding sites for CREB transcriptional factors (cAMP response-element binding proteins) and the H4TF-1 nuclear factor respectively. Mutation at these two sites could influence IL-18 expression and change the production of the cytokine [36]. In the IL-18 gene promoter transcription activity assay, after stimulation with PMA/ionomicin, low promoter activity was observed for A and C alleles at positions −607 and −137, respectively. In contrast, higher promoter activity was observed for C and G alleles at similar positions [36]. Theoretical higher frequencies of A alleles at position −607 and/or higher frequencies of C alleles at position −137 would confer some protective effect against the development of RA.

Up to date, the association between IL-18 promoter polymorphisms (−607A/C and −137C/G) and RA risk had been studied. However, Some studies indicated an association between IL-18 gene polymorphisms and RA risk, but some studies failed to find any association. In this paper, we updated previous studies of the association between IL-18 promoter polymorphisms (−607A/C and −137C/G) and RA risk. Previously, Pan et al. [37] and Chen et al. [38] have done a meta-analysis: the results showed no association between the IL-18 promoter polymorphisms (−607A/C and −137C/G) and RA risk in Caucasian populations or Asian populations, but Chinese population carrying A allele of −607C/A significant associated with decreased risk for RA was observed by Chen et al. [38]. In this meta-analysis, we included more studies and found that individuals carrying A allele or AA/AC genotype of −607A/C polymorphism was significantly associated with decreased risk for RA in allele model (OR = 0.778, 95 % CI = 0.633–0.955) and dominant model (OR = 0.618, 95 % CI = 0.466−0.819). But for −137C/G polymorphism, we failed to find any association with RA risk under all genetic models. In addition, subgroup analysis by ethnicity, we also observed significant association between the IL-18 −607A/C polymorphisms and RA risk in Asian populations, but not in Caucasian populations. However, we found no association between the IL-18 −137C/G polymorphism and RA risk in Asian populations or Caucasian populations.

The results of this meta-analysis should be explained with caution because of some limitations. Firstly, the sample size of the included studies was not large enough. For −607A/C polymorphisms in the subgroup analysis, the number of studies about Caucasian populations was only three studies; for −137C/G polymorphisms, only 1,331 cases and 1,468 controls were included. Secondly, the heterogeneity among studies was observed in our meta-analysis. We did meta-regression analyse of ethnicity, publication year, genotypic method and sample size, but failed to identify the sources of heterogeneity. It might be due to environmental backgrounds and other reasons. Thirdly, the outcomes were based on unadjusted estimate effects, and the effect of gene–gene and gene-environment interactions were not addressed in this meta-analysis. Fourthly, although the Begg’s funnel plot and Egger’s test showed no publication bias, bias of selection may have occurred because only studies in English or Chinese were selected.

In conclusion, this meta-analysis provide evidence that IL-18 −607A/C polymorphisms is associated with RA risk. Large-sample studies of different ethnic groups with carefully matched cases and controls are needed to clarify the role of the two promoter gene polymorphisms and RA risk in the future.