Association between rs7517847 and rs2201841 polymorphisms in IL-23 receptor gene and risk of ankylosing spondylitis: a meta-analysis

Search strategy and selection criteria

A comprehensive search of databases such as Pubmed, Embase, Medline, Cochrane, China National Knowledge Infrastructure (CNKI), VIP, Wanfang and China Biology Medicine disc (CBMdisc) databases was conducted. Searches included literature dated from database origin to September 2014, and the following key words were used: “IL-23” OR “interleukin-23”, “Ankylosing Spondylitis” OR “AS”, and “polymorphism” OR “polymorphisms”. In the CNKI, VIP, Wanfang and CBMdisc databases we searched for corresponding words in Chinese characters. The full search strategy for the Embase database is presented in Table 1. No language restrictions were used. A manual search for references beyond those in the above-mentioned databases was also implemented. For studies that did not describe genetic distribution data in detail, email correspondence with the main authors was used to complete the data. Titles and abstracts were independently screened by two authors to identify potentially related studies. Full-text versions of the identified studies were reviewed to select those that met the eligibility criteria. The identified studies were subjected to a final confirmation before inclusion in the meta-analysis.

Inclusion criteria

Investigations that met the eligibility criteria were included in the analysis, and any disagreements were resolved by discussion between the authors (Xu B and Ma JX). In cases where a consensus could not be reached, a third author was involved to make a final decision. Studies meeting the following criteria were included in the meta-analysis: (1) investigation evaluating the association between IL-23R rs7517847 or rs2201841 polymorphisms and AS susceptibility; (2) a case-control study or GWAS; (3) sufficiently available public data that could be extracted for further analysis, such as genotype distribution, odds ratio (OR) and 95% confidence interval (95% CI); and (4) a SNP distribution according to Hardy-Weinberg equilibrium (HWE) was included, which means high quality in design and conduct of genetic association studies. In cases where two studies examined the same or overlapping populations, the study with a larger sample size was included in the analysis.

Studies for which contact with the main author could not be reached to supply information about relevant data were excluded.

Data extraction and quality assessment

The relevant characteristics of the included investigations were identified and recorded by two authors, including the first author of the study, publication year, country, ethnicity of subjects, relevant SNPs, patient demographics, test method used for genotype, whether the genotype distribution was in accordance with Hardy-Weinberg Equilibrium (HWE) and source of the samples tested. To acquire precise results for the current study, email contact with the main author of the investigation was performed if the included study did not contain public data. Quality assessment of studies included in the meta-analysis was conducted by two authors using the Newcastle-Ottawa Scale (NOS) (Cota et al., 2013). Scores were given for subject selection (i.e., adequateness of the case definition, representativeness of the cases, selection of controls, and definition of controls) and the comparability of the groups (i.e., comparability of cases and controls on the basis of the design or analysis) as well as measurement of exposure (i.e., ascertainment of exposure, same method of ascertainment for cases and controls, and non-response rate). NOS scores ranged from 0 to 9. Studies with a NOS score ≥6 were considered to be high quality. Higgins I² was used to evaluate the heterogeneity of the investigations. Subgroup analysis by ethnicity was executed if the number of investigations in each ethnic group was two or more. Gender subgroup analysis could not be done because there was no relevant data. Sensitivity analysis was carried out by evaluating the overall results of meta-analysis when each study was removed to detect the stability of trials included.

Statistical analysis

The strength of the association between IL-23R rs7517847 and rs2201841 polymorphisms and AS susceptibility was evaluated using the Odds Ratio (OR). At the same time, precision was measured by 95% CI. A random-effects model was used in the current study. With respect to rs7517847 and rs2201841, the homozygote model, heterozygote model, recessive model, dominant model and allelic model were used to estimate susceptibility to AS in current study. Statistical analysis of the extracted data was carried out using STATA 12.0.

Search results

As a result of the search strategy presented in Fig. 1, 215 English-language studies and 208 Chinese-language studies were obtained. In addition, one study was added by searching through references (Zhu, Yang & Gao, 2009). Among the search results, 420 trials were excluded for the following reasons: (1) 116 were duplicate articles; (2) 281 did not discuss the association between IL-23R SNPs and AS in the title and abstract; (3) 20 did not discuss the association between rs7517847 and rs2201841 and AS in the full-text; (4) two contained genotype distribution data that were not publicly available, and email contact with the main authors could not be achieved; and (5) two studies assessed overlapping populations (Sáfrány et al., 2009; Szabo et al., 2013). For these studies, those that included a larger sample population were included in the analysis, as conducted by Sáfrány et al. (2009). Finally, four trials (Sáfrány et al., 2009; Dong et al., 2013; Zhu, Yang & Gao, 2009; Rueda et al., 2008) on rs7517847 and two trials (Sáfrány et al., 2009; Zhu, Yang & Gao, 2009) on rs2201841 were included in the current study. The included studies had case-control designs with 1,006 patients in the AS group and 1,190 people in the control group for rs7517847. For rs2201841, 322 patients were included in the AS group, and 255 people were included in the control group in the current study. The basic characteristics of included studies were listed in Table 2. Patients who were diagnosed with AS according to the New York modified criteria (Goei et al., 1985) and healthy people from Chinese, Hungarian and Spanish populations were included in the current study. All trials included in this meta-analysis were replication studies according to HWE. The NOS score of studies included in the meta-analysis ranged from 6 to 7, as presented in Table 3. Moreover, genotype distribution of both SNPs in the AS and control groups is presented in Table 4.

Overall results of the meta-analysis

ORs, P value, 95% CI and heterogeneity evaluation of meta-analysis for rs7517847 and rs2201841 are presented in Table 5 and Fig. 2 and in Table 6 and Fig. 3, respectively. The random effect model was applied to the studies. A comprehensive analysis of relevant SNPs was performed. Based on the meta-analysis results, there was a statistically significant association between the IL-23R rs7517847 polymorphism in the over-all population and AS susceptibility under two genetic models (homozygote model, i.e., GG vs. TT, were OR =0.76, 95% CI [0.59–0.99] and P = 0.038 and the allelic model, G vs. T, i.e., where OR = 0.88, 95% CI [0.78–0.99] and P = 0.032). The heterogeneity among studies under the two genetic models was I² = 0, P_H = 0.547 and I² = 0, P_H = 0.838, respectively, which indicated that there was no statistically significant heterogeneity among the trials for rs7517847. The analytical results presented in Table 6 implied that there was no statistically significant association between the IL-23R rs2201841 polymorphism and AS susceptibility under all of the genetic models tested.

Results of subgroup analysis by ethnicity

A subgroup analysis by ethnicity (Asians and Caucasians) was conducted for rs7517847, and contained two trials (one with 435 and 455 people in the case and controls groups and a second with 571 and 735 people in the case and control group, respectively). The random effect model was applied to each genetic model in both the Asian and Caucasian populations.

In accordance with the results of the subgroup analysis, there was no statistically significant association between the IL-23R rs7517847 polymorphism and AS in the Asian or Caucasian populations, as presented in Table 7.

Sensitivity analysis

A sensitivity analysis of trials investigating rs7517847 was carried out, aiming to evaluate the stability of the current study. After removing the study conducted by Sáfrány et al. (2009), we found that the OR for GG vs. TT changed from OR = 0.76 (95% CI [0.59–0.99], P = 0.038) to OR = 0.80 (95% CI [0.60–1.06] and P = 0.120) and the OR for G vs. T changed from OR = 0.88 (95% CI [0.78–0.99], P = 0.032) to OR = 0.88 (95% CI [0.77–1.01] and P = 0.078). After removing the study conducted by Dong et al. (2013), we found that the OR for GG vs. TT changed from OR = 0.76 (95% CI [0.59–0.99], P = 0.038) to OR = 0.82 (95% CI [0.60–1.11] and P = 0.203) and the OR for G vs. T changed from OR = 0.88 (95%CI [0.78–0.99], P = 0.032) to OR = 0.90 (95% CI [0.78–1.04] and P = 0.145). Excluding the other two studies, we found that the results of studies had not changed.