Introduction
Inflammatory bowel diseases (IBD), consisting of ulcerative colitis (UC) and Crohn’s disease (CD), are chronic inflammatory disorders of the gut that are probably the result of a dysregulated immune response to the gut microflora in genetically predisposed individuals, triggered by environment [1,2]. The annual incidence of UC varies from 0-19.2 per 100,000 in North America and from 0.6-24.3 per 100,000 in Europe, corresponding to a prevalence of 37.5-248.6 per 100,000 and 4.9-505 per 100,000, respectively. The incidence of CD is similar in western countries (0-20.2 per 100,000 in North America; 0.3-12.7 per 100,000 in Europe). The incidence and prevalence of IBD are increasing over time and in different regions around the world [3]. IBD are polygenic diseases for which up to 163 genes have been found to increase the risk of susceptibility [4-6]. Previous reports indicated several single nucleotide polymorphisms (SNPs) in different regions of the genome increasing risk of IBD. One of the most important non-human leukocyte antigen (HLA) common susceptibility alleles for autoimmunity is the 1858C/T SNP of protein tyrosine phosphatase non-receptor 22 (PTPN22) (rs2476601).
The PTPN22 gene encodes a protein tyrosine phosphatase, non-receptor type 22 (PTPN22), located on chromosome 1p13. This 110-kDa lymphoid-specific phosphatase (Lyp) plays a critical role as a negative regulator of T-cell activation by dephosphorylating T-cell receptor activation dependent kinases (Csk kinase) [7], expressed exclusively in immune cells; therefore, this gene may be associated with autoimmunity [8].
SNPs within the PTPN22 gene may affect the regulatory role of the Lyp. A non-synonymous SNP at position 1858 of PTPN22 gene (rs2476601) changes the amino acid from an arginine to a tryptophan and affects the ability of Lyp to interact with the Csk kinase, thus avoiding the formation of the complex and the resulting suppression of T-cell activation. In vitro experiments have shown that the PTPN22 1858T allelic variant binds less efficiently to Csk than does the C allele. This suggests that individuals lacking the C allele of PTPN22 may have a reduced capacity to downregulate T-cell responses and may therefore be more susceptible to autoimmunity. Several studies have investigated a potential association of PTPN22 C1858T polymorphism with various autoimmune disorders, including rheumatoid arthritis [9], systemic lupus erythematosus [10], Grave’s disease [11], and type 1 diabetes mellitus [12]. However, contradictory results have been published regarding the association of rs2476601 variant with either CD or UC. This inconsistency might be due to differences in sample size, patient ethnicity, or allele frequencies. Accordingly, given these results, we performed a meta-analysis to determine quantitatively the risk of CD and UC with the rs2476601 variant under an allelic, recessive, dominant and co-dominant model.
Materials and methods
Selection of eligible studies
This meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) criteria [13]. The eligible studies were obtained by searching online databases, MEDLINE and EMBASE. The following keywords were used for searching: (‘‘inflammatory bowel disease’’ OR ‘‘Crohn’s disease” OR “ulcerative colitis”) AND (“polymorphism* “OR “variant”) AND (‘‘protein tyrosine phosphatase nonreceptor’’ OR “PTPN22” OR “lymphoid protein tyrosine phosphatase”). The most recent research was performed on April 30th 2016, and there was no limitation on the research. The references of reviews and retrieved articles were also searched simultaneously to find additional eligible studies. Research was performed using the Entrez interface of the National Center for Biotechnology Information (ncbi) from the terminal of a Linux station (ubuntu 14.04.2) and through a graphical interface in EMBASE.
Data extraction
Relevant information from the articles were extracted from the studies independently by the first (AH) and second (CC) authors. Disagreements were resolved through discussion to reach consensus or were adjudicated by a third author. If genotype information was lacking, we tried to contact the corresponding authors in order to obtain the required data. If they did not provide data, those studies were excluded from our review. General characteristics (e.g. demography, genotyping method) of the included studies were extracted. We extracted the following information from each included study: first author’s name, publication year, sample size, source of controls, ethnicity, genotyping method, matching variables of controls with cases, age, sex distribution, and counts of alleles and genotypes in both cases and controls. We restricted our meta-analysis to candidate gene studies. The included studies had to meet the following criteria: 1) they concerned the association between rs2476601 and IBD; 2) they were case-control design or cohort studies; 3) they reported genotype frequencies or allele frequencies of the rs2476601 polymorphism. The exclusion criteria were: 1) review-articles, case reports and comments; and 2) duplicate publications.
Statistical analysis
The Hardy-Weinberg equilibrium (HWE) was examined in control groups using Fisher’s exact test. If the study was found not to be in HWE with a P value less than 0.05, it was considered to be disequilibrium. Allele frequencies of the PTPN22 rs2476601 polymorphisms in each of the studies were determined using the allele counting method. Allelic effect contrast was examined for the minor allele vs. the common allele. The genetic models evaluated for pooled ORs included allelic contrast (C vs. T), recessive models (CC vs. CT + TT), dominant models (CC + CT vs. TT), and the homozygote model (CC vs. TT). To evaluate the strength of association, the pooled ORs and their 95%CIs were determined for each study, and within- and between-study heterogeneity was assessed using Cochran’s Q statistic [14]. The heterogeneity test was used to assess the probability of the null hypothesis that all studies were evaluating the same effect. The random-effects model was used for meta-analysis when a significant Q statistic (P<0.10) indicated heterogeneity across studies, while the fixed-effect model was used when heterogeneity was not indicated. The fixed-effect model assumes that genetic factors have similar effects on disease susceptibility across all studies and that observed variations between studies are caused by chance alone [15]. The random-effect model assumes that studies show substantial diversity, and assesses both within-study sampling errors and between-study variances [16]. When study groups are homogenous, the two models are similar, but if this is not the case, the random effects model usually provides wider CIs than the fixed effects model. The random effects model is best used in the presence of significant between-study heterogeneity [16]. We quantified the effect of heterogeneity by using the recently developed I2 measure, where I2 = 100% × (Q − df)/Q, where df is degrees of freedom [17]. The I2 measure ranges between 0% and 100% and represents the proportion of inter-study variability attributable to heterogeneity rather than chance. I2 values of 25%, 50%, and 75% were defined as low, moderate, and high estimates, respectively. All statistical analyses were performed using the general package for meta-analysis (version 4.6-0, depends on R≥2.9.1).
Evaluation of publication bias
Finally, potential publication bias was assessed using a funnel plot and Egger’s test, which measures funnel plot asymmetry on a natural logarithm scale of ORs [18]. Funnel plot should be interpreted with caution, as they usually require a large number of studies.
Results
Search strategy
According to the search strategy (Fig. 1), 34 studies were initially included; 15 papers were excluded after reading the title and abstract, 1 paper was not relevant to rs2476601, while 1 paper [19] included three different populations, which were treated as three different studies. Two studies [20,21] that presented a significant deviation from Hardy–Weinberg equilibrium in controls (P=0.041 and P=0.010) were excluded from the analysis (Table 1). One article [22] was excluded on the basis of a T recessive test, as the TT genotype was not reported in the population of interest.
Figure 1 Flow chart showing selection of studies and specific reasons for exclusion from the study
HWE, Hardy-Weinberg equilibrium
Table 1 Hardy–Weinberg Equilibrium (HWE) deviation P-values in controls and cases
A total of 14 studies [9,23-33] were thus included in the analysis (Tables 2 and 3) involving a total of 13,356 controls, 8182 patients with CD, and 8656 with UC. The methodological quality of the studies were also evaluated (Table 4).
Table 2 Genotype and allelic distribution for rs2476601 in the selected studies
Table 3 Demographic and baseline characteristics of the selected studies
Table 4 Methodological quality of included studies
Quantitative analysis of PTPN22 rs2476601 and CD
They were 14 studies investigating the association between PTPN22 and CD, which included a total of 13,370 patients and 16,888 controls (Table 2). The presence of the C allele significantly increased the risk of having CDs (OR 1.28, 95%CI 1.17-1.40). We could not find significant heterogeneity between the studies (I2=37.5%, P=0.0835). The Funnel plot showed no asymmetry.
The CC genotype was compared with CT and TT assuming a recessive model for the C allele. The pooled OR was 1.34 (95%CI 1.20-1.49) and no significant between-study heterogeneity was found (I2=3.5%, P=0.4077) (Fig. 2). Assuming a dominant model for C allele, we could not find a significant risk for having CD (OR 1.14, 95%CI 0.72-1.8). The between-study heterogeneity was low (I2=14.5%, P=0.3094) (Fig. 2, Table 4).
Figure 2 Risk of Crohn’s disease associate with C allele, assuming an allelic, dominant, recessive, and co-dominant model
CD, Crohn’s disease; OR, odds ratio; 95%CI, 95% confidence interval
Quantitative analysis of PTPN22 rs2476601 and UC
Seven studies investigated the allelic risk of rs2476601, including a total number of 6971 patients with UC and 9715 controls. The results are summarized in a forest plot (Fig. 3). No significant association was found (OR 1.06, 05%CI 0.98-1.14), assuming a fixed effect model because of a low between-study heterogeneity (I2=32.2%, P=0.1064). No association was found assuming a dominant effect for C allele (OR 0.92, 95%CI 0.64-1.31), or a recessive model for C allele: OR 1.09, 95%CI 0.97-1.22, and OR 1.11, 95%CI 0.94-1.31, assuming a fixed effect and a random effect model respectively (I2=50%, P=0.0639). In addition, measuring the risk of carrying CC compared to TT allele on a co-dominant model, we could not find a significant increased risk of UC with either genotype (OR=0.84, 05%CI 0.55-1.28) assuming a fixed effect model.
Figure 3 Risk of ulcerative colitis, assuming an allelic, dominant, recessive and co-dominant model
CD, Crohn’s disease; OR, odds ratio; 95%CI, 95% confidence interval
Table 5 Summary statistics for Crohn’s disease (CD) and ulcerative colitis (UC)
Sensitivity analysis
We performed a sensitivity analysis for statistically significant results. For the association of the PTPN22 rs2476601 polymorphism and CD or UC susceptibility among the overall populations, the observed significant result was not materially altered by sequentially excluding each study.
Publication bias
Egger’s test indicated no significant publication bias in the allelic, dominant, recessive and co-dominant models for CD (P=0.48, P=0.75, P=0.78 and P=0.48, respectively). There was also no significant publication bias in the allelic, dominant, recessive and co-dominant models for UC (P=0.49, P=0.6, P=0.91 and P=0.49, respectively). The funnel plots were symmetrical for all tests.
Discussion
The incidence of IBD is increasing worldwide and the rising prevalence of immune diseases is thought to be influenced by genetic and environmental factors [1,2]. PTPN22 encodes the gatekeeper of T-cell receptor signaling, cytoplasmic Lyp and as such is a likely candidate risk factor for UC because of its role in down-regulation of T-cell activity [34].
So far, genome-wide association studies have found more than 163 genes that are associated with IBD in a Caucasian population from. Of these genes, 110 have been shown to be significantly associated with both UC and CD, but 23 showed risk effects that were UC specific and 30 were CD-only loci [5]. This suggests that there are different biological pathways in these two diseases that are well reflected by the different phenotypes and still need to be explored. There have been conflicting results, with some studies showing an increased risk associated with the T allele [19], whereas other studies, such as the one conducted by Silverberg et al [35], did not report any significantly increased risk. Most studies have failed to show a significant association and this could be due to a lack of power. A meta-analysis is a powerful tool that can improve the statistical performance by combining the results of multiple studies. Therefore, in the present study, we investigated whether the PTPN22 1858C/T polymorphism contributes to IBD susceptibility, using a meta-analysis approach.
We pooled data from 14 individual studies results to determine the effect of a non-synonymous SNP at position 1858 of the PTPN22 gene (rs2476601) on CD and UC, This is the first meta-analysis of candidate gene studies exploring the variant rs2476601 of PTPN22. We determined that individuals carrying minor allele C in rs2476601 have a greater risk of developing CD (OR 1.28, 95%CI 1.17-1.40) relative to those carrying the T allele, while carrying the C allele does not significantly change the risk of developing UC. Genotypic effects were also estimated for PTPN22 and the risk of CD was greater (OR 1.34, 95%CI 1.20-1.49) with CC genotype compared to CT or TT. In our study we did not detect significant between-study heterogeneity among the case-control studies included. Interestingly, the T allele was associated with a lower risk of CD, but an increased risk for other immune diseases, such as type 1 diabetes mellitus, or rheumatoid arthritis [10,12].
The rs2476601 polymorphism disrupts the interaction between Lck and Lyp, leading to reduced phosphorylation of Lyp, which ultimately contributes to gain-of-function inhibition of T-cell activation [36]. Despite the promising role for PTPN22 mutation in the risk of UC, we were unable to find a significant association in our meta-analysis. Interestingly, other polymorphisms, such as -1123G/C, located in the promoter region of PTPN22 have been shown to be correlated with other autoimmune diseases [37,38]. Another polymorphism, +788G/A in exon 10 of PTPN22, has been found to be associated with UC [22] in a Chinese population but remains to be further explored in the Caucasian population.
More recently, Spalinger et al [39] investigated the association between PTPN22 rs2476601 polymorphism and clinical features in 2028 IBD patients. They found that TT and CT genotypes were associated with less use of steroids and antibiotics and a lower prevalence of vitamin D and calcium deficiency in CD, which again indicates a protective role for the T allele. Interestingly, the authors found that TT and CT genotypes were significantly associated with an increased use of azathioprine and anti-tumor necrosis factor antibodies in UC, suggesting more severe clinical manifestations, and therefore opposite effects on disease severity compared to CD. However, in our study, we could not identify a significantly higher risk of UC with rs2476601 after pooling data from 6971 individuals. However, given the much smaller number of patients compared to those with CD, there may have been insufficient statistical power to detect a possible difference.
Several limitations in our study should be considered. First we could not perform a stratified analysis because of the lack of data on other ethnic groups. Most of the studies involved a Caucasian population. Gene polymorphism variation among different ethnicities was not explored. According to the study of Chen et al [22], the T allele was absent in rs2476601, suggesting that the protective effect on CD is exclusive to non-Asian ethnic groups. In addition, clinical subtypes of IBD, especially relating to the severity of the diseases, could not be taken into account due to insufficient data.
Nevertheless, the present meta-analysis also had several strengths. First, no publication biases were detected, indicating that the results may be unbiased. Second, a significant number of cases and controls were included in the current study by using an effective and efficient search strategy to increase the statistical power of the analysis. Third, to our knowledge, this is the first meta-analysis to assess the association between PTPN22 and IBD.
In conclusion, the present meta-analysis pooled both statistically significant and non-significant findings from individual studies to improve the statistical performance and generate a precise conclusion. The findings of this meta-analysis demonstrate that rs2476601 polymorphism is an important risk factor for CD but not for UC. However, larger and well-designed multicenter studies, particularly addressing the Asian population group, stratified by gene–gene and gene–environment interactions, are warranted to validate our findings.
What is already known:
-
The incidence and prevalence of inflammatory bowel diseases (IBD) are increasing over time and in different regions around the world, but their etiology remains unknown
-
One of the most important non-HLA common susceptibility alleles for autoimmunity is the 1858C/T single nucleotide polymorphism of protein tyrosine phosphatase non-receptor 22 (PTPN22) (rs2476601)
-
The PTPN22 gene is located on chromosome 1p13.3-p13.1 and encodes a lymphoid-specific phosphatase
-
The association between PTPN22 R620W polymorphism and IBD is inconsistent among candidate gene studies
What the new findings are:
-
Individuals carrying minor allele C in rs2476601 have a greater risk of developing Crohn’s disease relative to those carrying T allele, under an allelic and dominant model
-
No significant results were found concerning ulcerative colitis, suggesting different underlying biological pathways
-
No significant risk association could be identified under the recessive or co-dominant model for ulcerative colitis and Crohn’s disease. Studies were not heterogeneous and no publication bias has been identified