Contribution of NKX2-3 Polymorphisms to Inflammatory Bowel Diseases: A Meta-Analysis of 35358 subjects

Polymorphisms in NKX2-3 gene have been inconsistently associated with Crohn's disease (CD) and ulcerative colitis (UC). To generate large-scale evidence on whether NKX2-3 polymorphisms are associated with CD or UC susceptibility we have conducted a meta-analysis of 17 studies involving 17329 patients and 18029 controls. A significantly increased CD or UC risk was observed in persons carrying a G allele at rs10883365 polymorphism (A/G) compared with those with a A allele. (OR = 1.226, 95%CI: 1.177–1.277 and OR = 1.274, 95%CI: 1.175–1.382 respectively). In the subgroup analysis, a significantly increased CD risk was found in both Europeans and Asians. For rs11190140 polymorphism (C/T) and CD risk, the risk estimate for the allele contrast was OR = 1.201 (1.136–1.269). This meta-analysis provided a robust result that persons with a G or T allele may have a moderately increased risk of CD, and suggested that rs10883365 polymorphism was also a candidate gene polymorphism for UC susceptibility.

I nflammatory bowel diseases (IBDs) are chronic inflammatory disorders characterized by chronic relapsing inflammation of the gastrointestinal tract that affect 0.1% of Western populations, comprising two major forms, Crohn's disease (CD) and ulcerative colitis (UC) 1 . In Crohn's disease the inflammation is often transmural, whereas in ulcerative colitis the inflammation is typically confined to the mucosa. Additionally, Crohn's disease can be associated with intestinal granulomas, strictures, and fistulas, but these are not typical findings in ulcerative colitis. Although our understanding of disease pathogenesis remains incomplete, accumulating evidence suggests that that IBD is a complex, multifactorial disease partly determined by a genetic predisposition 2 . Strong familial aggregation, twin studies, and established genetic associations [3][4][5] . indicate that there is a genetic component to the disease susceptibility in IBD. Recently, sequence variations associated with IBD have been reported for several genes, including NOD2, IL23R, IRGM, ATG16L1, PTPN2, and NK2 transcription factor related and locus 3 (NKX2-3) [6][7][8][9][10][11] .
NKX2-3, located on 10q24, is a member of a family of genes that encodes transcription factors containing homeodomains and, therefore, is implicated in basic developmental functions. During development, NKX2-3 is expressed in midgut and hindgut mesoderm and spleen, as well as in pharyngeal endoderm 12,13 . The association between the NKX2-3 polymorphism and susceptibility of IBD was first reported in Caucasian patients 14 . After the first report of the association, several studies confirmed the association of tag-SNPs (rs10883365 and rs1190140) in the NKX2-3 gene with CD 15,16 as well as with UC in Caucasian or Asian populations [17][18][19] . However, several studies could not replicate the genetic association between IBD and NKX2-3 polymorphsims 15,20,21 .
Thus, a quantitative synthesis may help to provide clearer evidence on the association of such genetic polymorphisms with IBD. In the present study, we conducted a meta-analysis of all eligible studies to quantitatively assess the associations between three common polymorphisms (rs10883365 and rs11190140) in the NKX2-3 gene and IBD susceptibility.

Results
Characteristics of the included studies. The combined search yielded 75 references, of which 31 were duplicate studies, 9 were reviews, 4 were about cell studies, 8 were only with abstracts, 7 reported other mutations, 1 reported other disease. Finally, a total of 15 articles were finally included. Among them, one publication 15 contained data on two different subpopulations, one 16 included Wellcome Trust Case Control Consortium (WTCCC) samples and replication Crohn's disease (RCD) samples, and we treated them independently. In total, 17 studies comprising 17329 cases and 18029 controls were included in the present meta-analysis 11,15-28 . The 17 separate studies consisted of 13 European and 4 Asian. The distribution of genotypes in the control groups of all studies was in agreement with HWE except for 1 study 19 . Summaries of all included studies were summarized in Table 1, and the flow chart of study selection process was shown in Figure 1.
Quantitative synthesis. Crohn's disease. The summary of metaanalysis for the NKX2-3 polymorphisms with CD is shown in Table 2, Figure 2A and Supplementary Figure S1. Regarding rs10883365 polymorphism, the results of combined analyses comprising 8699 cases and 13540 controls revealed a significantly increased risk of CD in all genetic models. In addition, the OR was 1.481 (1.351-1.623) in carriers of two risk G alleles compared with non-risk allele carriers (GG vs AA), which was higher than the risk of one G allele carriers (GA vs AA, OR 5 1.141 (1.055-1.234), suggesting a dose-response with increasing number of the variant allele. In the subgroup analysis, significantly increased risks were found both among European and Asian population. No betweenstudy significant heterogeneity was observed in all genetic models. A total of 5484 patients and 4863 controls were investigated for rs11190140 variant, a significant association was found in all genetic models. (see Supplementary Fig. S1) Similar to rs10883365, the OR (OR 5 1.485, P , 0.001) in carriers of 2 risk alleles was higher than that (OR 5 1.155, P , 0.001) in those of 1 risk allele. No betweenstudy heterogeneity was detected in any genetic models of rs11190140 variant and CD risk.
Ulcerative colitis. Seven studies with 4996 UC patients and 5479 controls for rs10883365 polymorphism were investigated. Meta-analysis findings of associations between rs10883365 in NKX2-3 gene and the risk of UC were shown in Table 3 and Figure 2B  pooled OR still did not change at all, indicating that the results of this meta-analysis were high stable. (Table 2 and 3) In the cumulative meta-analysis, the pooled ORs tended to be stable and the associations tended toward significant associations with accumulation of more data over time between rs10883365 or rs11190140 variant and CD risk, as well as between rs10883365 and UC risk (see Fig. 3 and Supplementary Fig. S4).
Publication bias. Funnel plots and Egger's test were performed to assess publication bias. No publication bias was detected for rs10883365 polymorphism.  Figure S5, the shapes of the funnel plots did not indicate any evidence of obvious asymmetry for rs10883365 variant and CD or UC risk.

Discussion
Presently the mechanisms of the etiology and progression of IBD are far from clear. Several genes have been identified to be associated with IBD risk, including NOD2, NKX2-3 and IL-23. Recently, genome-wide association studies (GWAS) have identified SNPs implicating hundreds of replicated loci for common traits and becomes a powerful tool to detect the susceptibility genes in the IBD diseases 16,29,30 . Several GWAS and GWAS meta-analysis have provided strong evidences for the association between NKX2-3 single nucleotide polymorphisms (SNPs) (rs6584283 and rs4409764) and risk of IBDs 31-33 . However, two common variants (rs10883365 and rs11190140) in NKX2-3 gene were not included in the GWAS meta-analyses.
A meta-analysis can combine results from individual studies to overcome the limitation of small sample sizes and inadequate statistical power, and produce a single estimate of the major effect. Recently, accumulated meta-analysis has been performed to investigate the association of genetic variants with susceptibility to CD or UC. Polymorphisms in several genes, including ATG16L1 T300A 34 , TGF-a G308A 35 , MIF G173C 36 , OCTN1 C1672T 37 , CD14 C260T 38 and MDR1 C3435T 39 , were identified as risk factors of CD or UC. Patients with mutant allele of NOD1 rs6958571 40 and PPAR Pro12Ala 41 might have a decreased susceptibility to IBD. Additionally, some genetic variants were not association with CD or UC risk, such as MDR1 C1236T 39 , IL-10 G1082A 42 , and IL-18 A607C 43 . Therefore, we saw the need to perform pooled analyses with larger sample size by summarizing previous case-control or cohort studies in order to better understand the association between the NKX2-3 variants and IBD risk.
NKX2-3, located on chromosome 10q24, is predominantly expressed in mesoderm of midgut and hindgut during embryonic mouse development 12 . Postnatally, Nkx2-3 expression continues in gut mesenchyme and in spleen. In addition, mice lacking Nkx2-3 exhibit severe defects in gut development; primarily in the epithelium of the small intestine 44 . The perturbations of the gut tissue architecture lead to early postnatal death presumably due to digestive malfunctions. Moreover, analysis of Nkx2-3-deficient mice has revealed a critical role for NKX2-3 in spleen development and in establishing the correct environment for normal B cell development and T cell dependent immune response 45,46 . Recently, associations between the two common polymorphisms (rs10883365 and rs11190140) in NKX2-3 gene and susceptibility of CD or UC have been reported in several studies.   To the best of our knowledge, the present study involving 37039 subjects represents the first comprehensive meta-analysis investigating the association between NKX2-3 polymorphisms (rs10883365 and rs11190140) and IBD susceptibility. For the analysis of rs10883365 polymorphism, a significantly increased CD risk was observed in all genetic models. In the ethnicity-stratified analyses, significant association was found both in Asian and European populations. Similar results were found between the rs11190140 variant and risk of CD where a significant association was found European population (no Asian population reported). No between-study heterogeneity was observed in most genetic models. Sensitivity analysis indicated that when excluding studies departed from HWE, the pooled OR still did not change, demonstrating the results of this meta-analysis were stable. Since these two SNPs are close to each other, we use 1000 Genomes Pilot sequence data to identify whether these SNPs are in linkage disequilibrium (LD) (r 2 . 0.8). The results indicated that rs10883365 and rs11190140 are in perfect LD (r 2 5 1.0). CD and UC, as two major subtypes of IBD, are believed to share overlapping but distinct clinical and pathological features, and have great differences in genetic backgrounds 11 . Some genes, such as NOD2 and ATG16L1, were associated with CD, but not with UC 47,48 . However, recent GWAS meta-analysis identified 163 IBD loci that meet genome-wide significance thresholds, 50 of these have an indistinguishable effect size in UC and CD, including IL23R (rs11209026), IL10 (rs3024505) and MST1 (rs3197999). In the present meta-analysis, significant association between rs10883365 and risk of UC was found in all genetic models. When stratified by ethnicity, similar correlation was observed both in Asians and Europeans. No between-study heterogeneity existed in major genetic models. Sensitivity analysis showed no single study qualitatively changed the pooled ORs. Moreover, excluding studies departed from HWE, the pooled OR still did not change, demonstrating the results of this meta-analysis were stable. These results indicated that rs10883365 polymorphism in NKX2-3 gene may be significantly associated with both CD and UC.
By combining the data of individual studies, we increased the statistical power to detect subtle associations, however,several limitations should be considered in our meta-analysis. Only studies published in the English and Chinese language were included in this meta-analysis; therefore, publication bias may have occurred. In addition, this meta-analysis was designed to analyze single polymorphism, a haplotype analysis may have been more powerful for finding significant associations with UC and CD. Finally, geneenvironment interactions were not analyzed because of insufficient data.
Despite these limitations, our results still yield statistical results. Taken together, we expand previously individual studies on IBD by suggesting that NKX2-3 gene rs10883365 polymorphisms might contribute to the occurrence of both CD and UC, and suggested that persons with a T allele of rs1190140 variant might have a significantly increased risk of CD. Further studies or large case-control studies, especially studies emphasizing genotype-phenotype interaction should be performed to clarify possible roles of NKX2-3in IBD. Moreover, studies involved in NKX2-3 polymorphisms in different populations with larger sample size might need to be performed.

Methods
Search strategy. We searched PubMed and Embase to identify genetic association studies of the rs10883365 or rs11190140 polymorphism and IBD risks. Electronic searches were performed by using the following search terms: 'Inflammatory Bowel Disease' or 'IBD', 'Crohn's disease' or 'CD', 'ulcerative colitis' or 'UC', 'NKX2-3', 'rs10883365', or 'rs11190140', (the last search update was 1 September 2013). In addition, the reference lists of reviews and retrieved articles were checked by handsearch for additional potential studies. A study reported results from more than one population was considered as separate studies.
Inclusion and exclusion criteria. Studies were considered eligible if they had to meet the following criteria: (1) association between NKX2-3 polymorphisms and risk of IBD (CD or UC), (2) case-control or cohort studies. Studies were excluded for the following reasons: (1) articles only with an abstract and review articles (2) no control population, (3) studies considered overlapped with other studies.
Data extraction. Two authors extracted the following data independently from each of the eligible articles: first author, publication year, ethnicity of the participants involved (categorized as Europeans or Asians), number of cases and controls, information (age, mean age at diagnosis and sex) of cases and controls, and number of genotypes or allele frequency in cases and controls. Study authors were contacted for detailed data when there was insufficient information to determine the relationship between genetic polymorphism and IBD risk. Disagreements were resolved by discussion between the two authors.
Statistical analysis. Pooled odds ratios (ORs) and 95% confidence intervals (CIs) used to assess the strength of the association between IBD risk (CD or UC) and NKX2-3 polymorphisms (rs10883365 or rs11190140). The significance of the pooled  19,49,50 available for the association between rs11190140 variant and UC risk, we have performed meta-analysis of correlation between rs11190140 polymorphism and CD risk. Subgroup analysis was performed according to ethnicity. Between-study heterogeneity was evaluated by using the Chi-square based Q test and I2 test 51 . Heterogeneity was considered significant for P , 0.10, and a randomeffects model was used, otherwise, fixed-effects model was used. In addition, if heterogeneity was detected, Galbraith plots were used to visualize the impact of individual studies on the overall homogeneity, which spot the outliers as the possible major sources of heterogeneity 52,53 . Moreover, a meta-regression was used to delineate the major sources of between-study heterogeneity 54 .
Sensitivity analysis was carried out to evaluate the stability of the results after sequential removal of each study or by excluding those studies deviated from HWE. In addition, cumulative meta-analyses were carried out for each polymorphism through assortment of studies with publication time. Graphical evaluation of funnel plots and Egger's linear regression test were performed to assess publication bias 55 . If significant publication bias was detected, ORs and 95% CI would be adjusted by trim and fill methods 56 . All statistical analyses were performed by STATA software, version 12 (StataCorp LP, College Station, Texas).