Variants in the IL7RA gene confer susceptibility to multiple sclerosis in Caucasians: evidence based on 9734 cases and 10436 controls

Recently, numerous genome wide association studies (GWAS) and other case-control association studies examining the relationship between interleukin-7 receptor α chain (IL7RA) gene rs3194051, rs987107, rs11567686, and rs11567685 variants and multiple sclerosis (MS) risk have been conducted, but the conclusions have been inconsistent. The main objective of this meta-analysis was to more precisely explore the association of these four IL7RA variants with MS development. Twenty-seven eligible studies involving 9734 cases and 10436 controls were included in the present meta-analysis. Power calculation, publication bias, sensitivity analysis and cumulative meta-analysis were performed to derive a reliable conclusion. Our study indicated three IL7RA loci were significantly associated with increasing MS risk (rs3194051: recessive model: OR = 1.22, 95% CI 1.08–1.38; rs987107: recessive model: OR = 1.44, 95% CI 1.22–1.69; and rs11567686: dominant model: OR = 1.18, 95% CI 1.01–1.37). Additionally, IL7RA rs11567685 variants might not be related to MS development. In all, IL7RA locus polymorphisms could play an important role in the predisposition to MS, which could contribute to a better understanding the pathogenesis of multiple sclerosis.

the IL7RA rs11567686 and rs11567685 polymorphisms, which were suggested to have a potential association with susceptibility to MS 24 . In 2005, Zhang et al. described a significant association between IL7RA rs3194051 and rs987107 variants and an increased risk of MS observed in Swedish patients 23 . Subsequently, multiple studies were conducted to explore the impact of these IL7RA polymorphisms on the development and pathogenesis of MS in different ancestral groups; however, these studies provided conflicting results 17,21,22,[25][26][27] . To the best of our knowledge, no systematic review of such association has been carried out to contend with the issue of inconsistencies from different research studies. Therefore, we synthesized available evidence from all published studies regarding the relationship between IL7RA polymorphisms and MS and performed a meta-analysis to elucidate the association between these four single nucleotide polymorphisms (SNPs) and susceptibility to MS in Caucasians.

Materials and Methods
Search strategies. Two reviewers systematically searched literature from the PubMed, Embase, Google Scholar, China National Knowledge Infrastructure (CNKI) and MS Gene (http://www.msgene.org/) databases (up to June 14, 2016). We first explored the CNKI database, but no eligible studies could be retrieved (data not shown in Fig. 1). We then performed a search of English databases using the following keywords: (interleukin 7 receptor OR IL7R OR ILRA OR IL-7R-alpha OR CDW127 OR IL7RA OR CD127) AND (polymorphism OR mutation OR variant) AND "multiple sclerosis". Additional studies were manually examined from the references cited in the original literature. For case-control studies with overlapping data, the one with the largest sample size was included in this meta-analysis.
Study inclusion criteria. The following inclusion criteria were used for selecting suitable studies: (i) the study was on the association of IL7R rs3194051, rs987107, rs11567686 or rs11567685 polymorphisms with MS; (ii) the study used a case-control design; (iii) the study provided raw genotype data (such as GG, GA and AA genotypes) for calculating the odds radio with 95% confidence interval. Articles that did not meet the above inclusion criteria were excluded from our meta-analysis.
Data extraction. For each eligible study, the following data were extracted: (1) first author's name and publication year, (2) area and ethnicity of the participants, (3) the number of cases and controls or the distribution of genotypes (4) source of control and genotyping method, (5) age and gender information, and (6) type and diagnostic criteria for multiple sclerosis. Two reviewers independently completed this step and collected the data carefully. Any disagreement was resolved by a discussion with a third reviewer.
Quality assessment. The Newcastle-Ottawa Scale (NOS) criteria were used to assess the quality of retrieved studies, which included three aspects: object selection, comparability and exposure assessment 28 . Studies with at least six points were considered high quality studies.
Statistical analysis. The Hardy-Weinberg equilibrium (HWE) was performed for testing the genotype distribution of the control group within each included study, and a P-value greater than 0.05 meant that the study sample was representative of the population in the corresponding area.
(+means a statistically significant result; + + means the effect size is greater than +; − means a non-significant result) A P-value of less than 0.05 for the Z-test indicated statistical significance of the pooled ORs. We used the statistical software Power and Sample Size Calculation (PS) version 3.1.2 (http://biostat.mc.vanderbilt.edu/ wiki/Main/PowerSampleSize) 30 to perform power calculations regarding the association of each IL7RA variant with MS under the appropriate model, and a power value greater than 0.8 meant high statistical power in this meta-analysis. The I 2 test was conducted to estimate the between-study heterogeneity. The random-effect model (DerSimonian and Laird method) was used when significant heterogeneity existed among the studies (I 2 ≥ 50%); otherwise, the fixed-effect model (Mantel-Haenszel) was utilized. Sensitivity analysis was conducted by removing one study at a time, especially the one not in HWE, to evaluate the stability of the overall results. Additionally, a cumulative meta-analysis sorted upon sample size was carried out to explain precisely the association of IL7RA rs3194051, rs987107 and rs11567686 polymorphisms with MS risk under their special genetic models, which updated the pooled results every time a new study sample was added. Publication bias was estimated by Begg's funnel plot and Egger's tests among the eligible studies, and a P-value > 0.05 suggested no apparent bias. The software STATA 14.0 (Stata Corporation College Station, Texas, USA) was used for all statistical analyses.

Results
Literature search results. Figure 1 describes the procedure of the literature search and the study selection in this meta-analysis. Following the initial search strategy, 303 potential publications were identified. Among them, 38 were found to be duplicates. After these duplicates were removed, we obtained 265 articles, which included 56 that were related to other diseases, 112 concerning unrelated loci and 7 articles whose studies did not utilize a case-control design. Therefore, these 175 articles were excluded and 90 articles remained. Subsequently, we further screened these remaining articles (n = 90) by removing the reviews (n = 60), animal studies (n = 13) and those studies with missing partial genotype data (n = 2). Finally, 15 articles including 27 eligible studies were included in this meta-analysis 17, 21-27, 31-37 . Among these 27 individual studies, 8, 6, 5, and 8 studies were linked to IL7RA rs3194051, rs987107, rs11567686, and rs11567685 polymorphisms, respectively. Table 2 provides the number of cases and controls, the number of genotypes of IL7RA rs3194051, rs987107, rs11567686, and rs11567685 loci, HWE and power analysis for each included study. From this table, it can be determined that the genotypic frequency distribution of the control group was consistent with HWE in all eligible studies, except for three (one reported by Zhang et al. 23 and two by Haj et al. 25 ). Further, the statistical power for all included studies under the applicable model ranged from 0.05 to 0.98. Table 3 describes the main characteristics of all the studies included in this meta-analysis, which could be divided into some diverse subgroups, regardless of some studies whose data might not be available. The genotyping method was divided into three subgroups: 'PCR' 23 , 'RT-PCR' 17,22,26,33,34,37 , 'PCR-RFLP' 21,24,25,27,31,32 , and the diagnostic criteria were also classified into three subgroups: 'Poser' 23 groups, and 0 to 19.8% in the PP groups. In addition, the NOS results suggested that all eligible studies in this meta-analysis were of high quality because their scores were equal to or greater than six points.

IL7RA rs3194051 polymorphism and MS risk.
Eight studies containing a total of 7292 MS patients and 8142 healthy controls were included in this meta-analysis to investigate the association between the rs3194051 polymorphism and MS risk. The estimated OR1, OR2, and OR3 were 1.21 (95% CI: 1.06-1.37), 0.97 (95% CI: 0.90-1.03), and 1.17 (95% CI: 1.02-1.33), respectively, suggesting a recessive genetic effect (GG vs. GA + AA) of MS risk allele G. A fixed effect model was conducted because only moderate heterogeneity (I 2 = 43%) existed in the recessive model. A statistically significant association was observed between the rs3194051 SNP and the susceptibility to MS (OR = 1.22, 95% CI: 1.08-1.38) ( Table 4 and Fig. 2A). In addition, the statistical power calculation based on the inclusion of sample size gave a value of 0.99, which indicated powerful evidence for the conclusion of this significant association.    (Table 4 and Fig. 2D).

Sensitivity analysis and cumulative meta-analysis.
A leave-one-out sensitivity analysis showed that the pooled ORs were not significantly changed for rs3194051, rs987107 and rs11567686 variants when all the included studies, containing those three studies distracted from HWE (investigated by Zhang et al. 23 and Haj et al. 25 ), were excluded one by one. This indicated that our results were robust and reliable (data not shown).
In the cumulative meta-analysis sorted by sample size, the pooled results detected a dynamic tendency of increased association between the minor variants of these three loci and the risk of MS under their most applicable genetic models, which confirmed our earlier conclusion. As an example, Fig. 3 described a tendency of increased association between the rs987107 polymorphism and MS risk in the recessive model. The combined ORs were not significantly fluctuated from accumulating each new study sample, which was also consistent with the findings of the sensitivity analysis.

Publication bias. In the present meta-analysis publication bias was estimated by Begg's funnel plot and
Egger's quantitative test. From the shape of funnel plots, we did not observe any asymmetric signal under all analyzed models (Fig. 4 illustrates no publication bias for the association of the rs987107 polymorphism with MS  Table 4. Meta-analysis of IL7RA polymorphisms on MS. Note: bold: significant P-value (<0.05); bold * : marginal association (0.05 < P-value < 0.1); # Suggested model.
risk.). The Egger's test also did not display any evidence of obvious publication bias for the association of these SNPs with MS risk (Table 4).

Discussion
Multiple sclerosis, an immune mediated disease in which T cells play an important role, is the chronic inflammatory neurologic disorder of the CNS affecting young adults, especially women [38][39][40] . There is growing evidence that genetic factors might play vital roles in MS development. Human leukocyte antigen (HLA), for example, has been widely reported to have a strong effect on MS 11,41,42 . IL7RA, on the other hand, serves as the first non-HLA gene that was also determined to have an association with MS susceptibility 17,21,23,43 . The present study is the first meta-analysis on the relationship between IL7RA variants and development of MS. IL7R functions as a significant pleiotropic receptor for the signaling pathway of IL7 in autoimmune disease 44 . IL7 interacts with the IL7R common gamma chain (namely CD132) and its alpha chain (namely IL7RA or CD127), forming the signaling complex in the IL7 cascade. The IL7/IL7R interaction is vital to the survival, proliferation, and differentiation of T-cells, especially CD4+ T-cells, which exist in the inflammatory lesions of the people with multiple sclerosis [45][46][47] . Thus, there is no doubt that an experimental investigation of the association of polymorphisms in the IL7RA gene with MS could help us to better understand the pathogenic mechanisms and develop molecularly targeted agents for MS treatment. Previous case-control studies have reported on the association between multiple IL7RA variants and the risk of MS, but the conclusions were inconsistent because of low statistical power, small sample size, or the complex gene-gene and gene-environment interactions involved in the disease. Therefore, to evaluate the function of IL7RA polymorphisms on MS more precisely, we conducted this meta-analysis, which increased statistical power by pooling the available data from individually published studies.
We collected eight studies on the rs3194051 variant and six studies on the rs987107 polymorphism; the baseline characteristics of these studies on genetic polymorphism of these two loci shared considerable similarities. The combined results from our meta-analysis indicated that both of these IL7RA polymorphisms were associated with increasing MS risk with high statistical power; this finding is well-matched with the conclusions of two individual studies investigated by Zhang et al. 23 and Lundmark et al. 22 . Among the other studies 17,26,[34][35][36] that suggested no obvious association between these two polymorphisms and MS risk, one involving rs3194051 and two involving rs987107 conducted by Gregory et al. 17 and Jäger et al. 26 , respectively, was actually shown to have a significant trend of developing MS since their individual confidence intervals were just slightly across 1 ( Fig. 2A,B). Moreover, the studies by O'Doherty et al. 36 on both polymorphisms and the study by Kallio 34 on the rs3194051 minor allele only suggested no relationship with MS but had relatively small sample sizes. Further, the distinct main features of cases and controls in individual studies, such as the different degrees of disease development among cases and the dissimilar genotype distributions in different geographical regions, might explain the lack of a significant association between the rs3194051 polymorphism and MS in the other two studies by Akkad et al. 27 and Bahlo et al. 35 . Thus, our conclusions of the association of both rs3194051 and rs987107 polymorphisms with MS risk should be reliable, especially with the high statistical power calculated in this meta-analysis.
We included five studies in this meta-analysis that investigated the relationship between the rs11567686 polymorphism and MS risk. The pooled results under the dominant model concluded that the rs11567686 minor alleles were statistically significantly associated with the susceptibility to MS. This conclusion would be reliable since it was also derived from a combined odds ratio with high statistical power. Though all the included studies individually indicated that rs11567686 polymorphism was not related with increasing MS risk, two studies 25,32 , especially the one investigated by Hoe et al. 32 with the largest sample size, showed a possibility of marginal association between this polymorphism and the disease (as shown in Fig. 2C). In addition, the association was also found in the stratified subgroup of SP + PP MS patients when compared with healthy controls (P < 0.05) from two reports 25,31 . Thus, the significant finding that the combined result was different from that of any of the included studies reflected the advantage of meta-analysis, by which we might properly evaluate the real genetic effect on disease development with greater statistical power through pooling all samples or synthesizing overall data available in previous studies.
In this retrospective analysis, eight studies were included to explore the effect of the rs11567685 variant on multiple sclerosis. For this association study we could not derive its most applicable genetic model. Thus, we utilized six potential genetic models to explore the association between the rs11567685 polymorphism and the predisposition to MS. Our results under all models suggested no relationship between this polymorphism and MS development, which was in agreement with the results from seven studies 24,25,27,[31][32][33]37 . Only one study conducted by Ibayyan et al. 21 indicated a association between the rs11567685 polymorphism and MS development; it is possible that the contradictory conclusion from this study might be attributed to its small sample size. Additionally, low between-study heterogeneity and a lack of publication bias obtained from this analysis further suggests the credibility of our conclusion.
There are four advantages in our study. First, meta-analysis has been recognized as an effective method to address a wide variety of clinical questions in evidence-based medicine by combining the results of multiple previously reported quantitative studies. To the best of our knowledge, this is the first meta-analysis on the issue of IL7RA variants with susceptibility to MS. Second, our meta-analysis had a relatively large sample size and strong statistical power, which helped to make the conclusion more convincing. Third, no obvious between-study heterogeneity and publication bias were observed in this retrospective analysis. Fourth, the results from both sensitivity analysis and cumulative meta-analysis confirmed the robustness of our conclusions.
The findings from the studies reviewed in this analysis should be interpreted with caution for several reasons. First, MS is a multifactorial disease and many other factors including age, gender, control of source, latitude, genotyping methods and gene-gene interactions might contribute to its susceptibility, but due to the insufficient data, we could not perform corresponding subgroup and stratified analyses to further explore in-depth reasons for MS pathogenesis. Second, our results might not be generalizable to other ethnicities because all of the included studies involved Caucasians. Third, a language bias may have existed because this meta-analysis only included English articles due to database limitations. Despite the above limitations, the present study is the first comprehensive meta-analysis with high statistical power that helps to expand our knowledge about the molecular biology and functional significance of IL7RA polymorphisms and the relationship with MS susceptibility.
Taken together, our meta-analysis indicated that IL7RA rs3194051, rs987107 and rs11567686 variants might contribute to the genetic susceptibility of MS, while the rs11567685 polymorphism had no effect on multiple sclerosis. While these results could provide a better understanding of MS pathogenesis, future well-designed studies with large sample sizes, gene-gene and gene-environment interactions are needed to confirm our present conclusions.