SEARCH

SEARCH BY CITATION

Keywords:

  • single nucleotide polymorphism;
  • meta-analysis;
  • shared genetic risk;
  • ulcerative colitis;
  • Crohn's disease

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Supporting Information

Background:

Both ulcerative colitis (UC) and Crohn's disease (CD) have a complex etiology involving multiple genetic and environmental factors. Many genome-wide association studies (GWAS) and subsequent replication studies revealed that both diseases share some of the susceptibility loci; however, common genetic factors for both diseases are not fully elucidated. This study is aimed to identify the common genetic factors for CD and UC by a meta-analysis of published studies.

Methods:

We first reviewed the 10 GWAS for CD to select candidate single nucleotide polymorphisms (SNPs). Next, we performed a PubMed literature search up to June 30, 2010 and carried out a systemic review of published studies that examined the association of CD susceptibility loci in UC patients. Meta-analysis was carried out using the inverse variance-weighted method or the DerSimonian-Laird method after estimating the heterogeneity among the studies. The data for highly linked SNPs were combined. Finally, we performed a meta-analysis of 43 published studies in 45 SNPs located at 33 loci by using a total of 4852 to 31,125 subjects.

Results:

We confirmed the association of 17 reported common susceptibility loci. Moreover, we found associations at eight additional loci: GCKR, ATG16L1, CDKAL1, ZNF365, LRRK2-MUC19, C13orf31, PTPN2, and SBNO2. The genetic risk of each locus was modest (odds ratios ranged from 1.05–1.22) except IL23R.

Conclusions:

These results indicate that CD and UC share many susceptibility loci with small genetic effect. Our data provide further understanding of the common pathogenesis between CD and UC. (Inflamm Bowel Dis 2011)

Ulcerative colitis (UC) and Crohn's disease (CD), the two most common forms of inflammatory bowel disease (IBD), have a complex etiology involving multiple genetic and environmental factors. Family and twin studies have clearly indicated the involvement of genetic factors in the development of both diseases.1 Moreover, UC and CD exist in the same family with higher frequency than the co-occurrence by chance alone, suggesting an etiological relationship between the two diseases.2, 3 Since the chronic relapsing intestinal inflammation induced by the dysregulated mucosal immune response to commensal enteric bacteria is one of the common pathogenesis of CD and UC, it is important to understand the shared genetic factors for both diseases.

Recent genome-wide association studies (GWAS) for CD4–13 have identified more than 30 susceptibility loci and provided new insights into the immunopathogenesis of this disease, implicating an important role of genes of the innate and adaptive immune systems for disease occurrence.14 Similarly, several GWAS for UC15–20 have identified more than 10 susceptibility loci. A comparison of the results of these studies and additional association studies has identified 18 common susceptibility loci between CD and UC, including IL23R, JAK2, STAT3, BSN-MST1, CCNY-CREM, KIF21B, NKX2-3, IL12B, ORMDL3, ICOSLG, LOC441108, IRGM, CCR6, TNFSF15, 5p13, 6p21, 7p12, and 21q21.16, 19–25 However, considering the strong heritability of both diseases, several common genetic factors may not have been found yet and meta-analysis of published studies is one approach by which these factors may be identified. Nevertheless, to our knowledge, only a handful of meta-analysis for common susceptibility loci between UC and CD have been performed, most notably for NOD2, PTPN22, ATG16L1, and IRGM.26–31 Therefore, we performed a comprehensive meta-analysis of published studies that examined the association of CD susceptibility loci in UC patients to clarify common genetic factors for both diseases.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Supporting Information

Single Nucleotide Polymorphism (SNP) Selection for a Literature Search

We reviewed the literature of 10 GWAS for CD including meta-analyses4–13 published before June 30, 2010. Initially, we selected 62 SNPs for the literature search based on the following criteria: 1) SNPs showed a significant level of overall P-value less than 5 × 10−7 in an initial GWAS for CD; and 2) located at non-MHC region because of the broad and strong linkage disequilibrium across the MHC region (Supporting Information Table 1).

Literature search strategy and study selection criteria

We performed a PubMed literature search (National Center for Biotechnology Information [NCBI]; http://www.ncbi.nlm.nih.gov/pubmed/) up to June 30, 2010 using the following terms: (ulcerative colitis or inflammatory bowel disease) and (polymorphism* or variant* or loci or locus). References from the selected publications were manually scanned to identify other relevant studies. Studies were included if: 1) they were case–control studies for Caucasian UC; 2) they included at least 100 UC cases; 3) they were published in English; 4) they examined the selected SNPs or the highly linked SNPs with the selected ones (r2 ≥ 0.95 in the HapMap Southern Utah residents of European descent [CEU] samples [release #27, build 36]); and 5) they provided enough data to calculate odds ratios (ORs) and 95% confidence intervals (CIs). For publications using overlapping samples, we discarded the smaller dataset (13 studies). The literature search and data extraction were conducted by two authors (K.A. and J.U.). Disagreement over eligibility was resolved by a detailed discussion after review by one additional author (T.M.). Details of this search strategy are shown in Figure 1. Finally, a total of 43 articles16, 19–25, 27, 29, 30, 32–63 were included in the meta-analysis (Table 1).

thumbnail image

Figure 1. Flowchart of search strategy for meta-analysis.

Download figure to PowerPoint

Table 1. Studies Included in the Meta-analysis
 StudyReferenceYearPopulationCaseControl
1Ogura322001USA182287
2Cuthbert332002UK566290
3Esters342004Belgium173165
4Büning352005Hungary128208
5Martín362005Spain544812
6Waller372006UK512750
7Oostenbrug272006Netherlands207276
8Crawford382007USA172104
9Cucchiara392007Italy186347
10Tremelling402007UK and Scotland9751345
11Büning_1412007Germany and Hungary296707
12Cummings422007UK6471134
13Glas432007Germany4561381
14Economou442007Greece180100
15Büning_2452007German and Hungary294845
16Roberts462007New Zealand466591
17Glas472008Germany5071615
18Lappalainen482008Finnland459292
19Márquez492008Spain363546
20Franke152008Germany11031817
21Fisher212008UK17401492
22Lakatos502008Hungary149149
23Okazaki512008Canada117310
24Roberts522008New Zealand475576
25Fowler532008Australia5431244
26Weersma_1542009Netherlands11201350
27Anderson232009UK25273028
28Silverberg162009USA and Canada10522571
29Weersma_2242009Belgium and Netherlands14421045
30Einarsdottir552009Sweden455280
31Glas562009Germany4761503
32Newman572009Canada4021005
33Palomino-Morales292009Spain425572
34Márquez_1302009Spain368745
35Márquez_2582009Spain405800
36Törkvist592010Sweden9351460
37Festen252010Netherlands14551902
38Sventoraityte602010Lithuania123186
39Lacher612010Germany132253
40Cénit622010Spain4421692
41Franke192010Germany10431703
42McGovern_GWAS1202010USA7232880
 McGovern_GWAS2202010Sweden9481408
 McGovern_GWAS3202010USA and Canada10222503
 McGovern_Replication1202010Italy993826
 McGovern_Replication2202010Netherlands1016754
43Perdigones632010Spain6621361

Meta-analysis

We assessed heterogeneity across the studies using Cochran's Q test and I2 statistics. P-value > 0.10 and I2 statistics < 25% indicated a lack of heterogeneity.64 If there was no heterogeneity among the studies, meta-analysis was carried out using the inverse variance-weighted method. This method is a fixed-effect model based on the assumption that the true OR of all studies is the same and no interstudy variance exist. When heterogeneity was present, we used the DerSimonian-Laird method. This method is a random-effect model which considers interstudy variance to estimate the combined OR. Publication bias was investigated by funnel plot and evaluated using Egger's test.65 Funnel plot is a scatterplot which displays the OR of each study on the X axis against sample size on the Y axis. If there is no publication bias, OR will be distributed symmetrically and its variation may be smaller in larger studies. The degree of symmetry of funnel plot was estimated by Egger's test. We considered the evidence of significant publication bias as an obvious asymmetry of funnel plot and Egger's P-value < 0.05. All statistical analyses were undertaken using R (http://www.r-project.org/).

We basically used reported ORs and 95% CIs of the published studies to perform meta-analysis. Since 15 out of 43 articles did not report OR or 95% CI, we calculated OR and 95% CI of each SNP using genotype data in eight studies,21, 27, 33, 35, 38, 39, 53, 61 sample size and minor allele frequency (MAF) in three studies,32, 34, 37P-value and OR in three studies,16, 20, 59 and P-value and MAF in one study.48 Among the 62 SNPs initially selected, we excluded seven SNPs (rs10801047 [1q31], rs1002922 [5p13], rs10512734 [5p13], rs1373692 [5p13], rs3810936 [TNFSF15], rs7848647 [TNFSF15], and rs5743289 [NOD2/CARD15]) because these SNPs had not been studied in at least two studies. In addition, the data of SNPs in ATG16L1 (rs2241880, rs10210302, and rs3828309), BSN-MST1 (rs9858542 and rs3197999), 5p13 (rs4613763 and rs17234657), IRGM (rs13361189, rs1000113, and rs11747270), TNFSF15 (rs6478108 and rs426389), NKX2-3 (rs11190140 and rs10883365), and NOD2/CARD15 (rs17221417, rs2066843, and rs2076756) were combined because these SNPs were in high linkage disequilibrium with each other (r2 ≥ 0.95) in the HapMap CEU samples. Finally, we performed a meta-analysis for 45 SNPs located at 33 loci by using a total of 4852 to 31,125 subjects. For an easy understanding of the risk direction, we calculated the OR and 95% CI of each SNP according to the risk allele in the GWAS for CD. A P-value less than 0.0015 (0.05/33) was considered statistically significant after applying Bonferroni correction.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Supporting Information

We found evidence of heterogeneity among the studies for 19 SNPs: rs2476601 (PTPN22), rs2274910 (ITLN1), rs2241880-rs10210302-rs3828309 (ATG16L1), rs4613763-rs17234657 (5p13), rs2188962 (LOC441108), rs10077785 (LOC441108), rs4958847 (IRGM), rs6908425 (CDKAL1), rs1456893 (7p12), rs1551398 (8q24), rs6478108-rs4263839 (TNFSF15), rs17582416 (CCNY-CREM), rs10995271 (ZNF365), rs10761659 (ZNF365), rs7927894 (C11orf30), rs2872507 (ORMDL3), rs2542151 (PTPN2), rs1736135 (21q21), and rs762421 (ICOSLG). Therefore, the pooled ORs and 95% CIs were calculated using a random-effect model in these variants. We found a significant publication bias at rs9292777 on 5p13 locus (Egger's P = 0.02) and excluded this SNP from the analysis.

Among the 45 SNPs included in the meta-analysis, 35 SNPs located at 30 loci were investigated by more than five studies. Among the 33 loci examined in this study, we found significant associations with UC in 14 loci and nominal associations (P < 0.05) in 11 loci. We confirmed the associations of 17 susceptibility loci which are commonly associated with both CD and UC in the previous study20: IL23R, KIF21B, BSN-MST1, 5p13, LOC441108, IRGM, IL12B, CCR6, 7p12, JAK2, TNFSF15, CCNY-CREM, NKX2-3, ORMDL3, STAT3, 21q21, and ICOSLG (Supporting Information Table 2). Moreover, we found associations with UC in eight additional loci (Table 2): GCKR (rs780094, P = 2.47 × 10−2, OR 1.05), ATG16L1 (rs2241880-rs10210302-rs3828309, P = 4.70 × 10−2, OR 1.05), CDKAL1 (rs6908425, P = 7.68 × 10−3, OR 1.10), ZNF365 (rs10761659, P = 4.67 × 10−4, OR 1.14), LRRK2-MUC19 (rs11175593, P = 1.54 × 10−2, OR 1.21), C13orf31 (rs3764147, P = 1.80 × 10−2, OR 1.07), PTPN2 (rs2542151, P = 2.49 × 10−2, OR 1.08), and SBNO2 (rs4807569, P = 1.72 × 10−2, OR 1.06). For all loci showing association, the directions of risk alleles for UC were all the same as those for CD. The OR of IL23R locus was relatively high (rs11209026, OR 1.62, 95% CI: 1.48–1.77), whereas ORs of other loci were modest ranged from 1.05–1.22.

Table 2. Results of Meta-analysis for Eight Additionally Identified Common Susceptibility Loci for CD and UC
 Allele* [1/2]StudyNumberRAFOR (95%CI)CombinedHeterogeneityPublication Bias
CaseControlCaseControlPOR (95% CI)PI2 StatisticsP
  • *

    Allele “1” denotes the reported risk allele.

  • †OR and 95% CI were calculated using the random-effect model because of the heterogeneity among the studies.

  • a

    rs1893217 is absolutely linked with rs2542151 (r2 = 1.0).

  • b

    rs2024092 is absolutely linked with rs4807569 (r2 = 1.0).

  • RAF, risk allele frequency; OR, odds ratio; CI, confidence interval; NA not applicable.

GCKR            
rs780094T/CAnderson (2009)246440020.400.381.07(0.99-1.16)2.47E-021.05(1.00-1.09)0.4800.40
  Franke (2010)104317030.420.401.10(0.99-1.23)     
  McGovern (2010) GWAS#172328801.03(0.88-1.20)     
  McGovern (2010) GWAS#294814081.00(0.94-1.07)     
  McGovern (2010) GWAS#3102225031.08(0.96-1.21)     
  Total620012496        
ATG16L1            
rs2241880G/ABüning_1 (2007)2967070.520.511.10(0.89-1.35)4.70E-021.05(1.00-1.10)0.110.310.43
rs10210302T/CRoberts (2007)4665910.510.501.05(0.87-1.25)     
rs3828309G/AGlas (2008)50716150.550.521.15(0.98-1.36)     
  Lappalainen (2008)4591900.460.470.96(0.75-1.23)     
  Franke (2008)107717930.550.531.19(1.01-1.41)     
  Fisher (2008)173914910.540.521.08(0.97-1.20)     
  Lakatos (2008)1491490.540.501.26(0.91-1.74)     
  Okazaki (2008)1173100.500.481.02(0.61-1.68)     
  Fowler (2008)54312440.480.510.87(0.75-1.01)     
  Newman (2009)40210051.19(1.00-1.41)     
  Weersma_1 (2009)112013500.550.560.95(0.84-1.08)     
  Palomino—Morales (2009)4146660.540.511.10(0.92-1.32)     
  Márquez_1 (2009)3687450.510.530.93(0.78-1.12)     
  Sventoraityte (2010)1231860.530.481.26(0.91-1.75)     
  McGovern (2010) GWAS#172328801.08(0.95-1.22)     
  McGovern (2010) GWAS#294814080.91(0.79-1.05)     
  McGovern (2010) GWAS#3102225031.08(0.95-1.22)     
  McGovern (2010) Rep#19938261.08(0.94-1.23)     
  Total1146619659        
CDKAL1            
rs6908425C/TFranke (2008)110217940.810.791.18(1.01-1.39)7.68E-031.10(1.02-1.18)0.130.390.33
  Anderson (2009)245340340.800.771.18(1.08-1.29)     
  Weersma_2 (2009)144210450.810.781.18(0.99-1.41)     
  McGovern (2010) GWAS#172328801.05(0.90-1.22)     
  McGovern (2010) GWAS#294814081.03(0.91-1.16)     
  McGovern (2010) GWAS#3102225031.11(0.95-1.30)     
  McGovern (2010) Rep#19938260.91(0.76-1.09)     
  Total868314490        
ZNF365            
rs10995271C/GTörkvist (2010)93514601.03(0.89-1.18)1.37E-011.07(0.97-1.17)0.020.680.26
  Franke (2010)104317030.440.401.19(1.07-1.33)     
  McGovern (2010) Rep#19938261.09(0.95-1.24)     
  McGovern (2010) Rep#210167541.00(0.96-1.05)     
  Total39874743        
rs10761659G/AFranke (2008)108817750.580.541.10(1.02-1.19)4.67E-041.14(1.05-1.23)0.240.28NA
  Fisher (2008)180715490.570.541.19(1.07-1.31)     
  Total28953324        
LRRK2—MUC19           
rs11175593T/CAnderson (2009)302611320.020.011.31(0.99-1.74)1.54E-021.21(1.03-1.41)0.7000.08
  Törkvist (2010)93514601.11(0.68-1.80)     
  Franke (2010)104317030.020.021.18(0.83-1.70)     
  McGovern (2010) Rep#19938261.31(0.97-1.76)     
  McGovern (2010) Rep#210167540.94(0.62-1.43)     
  Total70135875        
C13orf31            
rs3764147G/AAnderson (2009)242440170.220.211.07(0.98-1.18)1.80E-021.07(1.01-1.13)0.390.030.78
  Törkvist (2010)93514601.22(1.04-1.42)     
  Franke (2010)104317030.250.251.02(0.89-1.15)     
  McGovern (2010) Rep#19938261.04(0.90-1.20)     
  McGovern (2010) Rep#210167541.02(0.89-1.16)     
  Total64118760        
PTPN2            
rs2542151G/TFranke (2008)100517790.190.151.33(1.11-1.59)2.49E-021.08(1.00-1.16)0.140.370.06
  Fisher (2008)173514880.170.171.07(0.93-1.22)     
  McGovern (2010) GWAS#1a72328801.14(0.95-1.36)     
  McGovern (2010) GWAS#2a94814081.04(0.90-1.20)     
  McGovern (2010) GWAS#3a102225031.03(0.85-1.24)     
  McGovern (2010) Rep#19938261.00(0.92-1.09)     
  McGovern (2010) Rep#210167541.13(0.94-1.35)     
  Total744211638        
SBNO2            
rs4807569C/AAnderson (2009)242540470.220.201.10(1.00-1.20)1.72E-021.06(1.01-1.12)0.5700.85
  Franke (2010)104317030.250.241.03(0.90-1.17)     
  McGovern (2010) GWAS#1b72328801.00(0.85-1.18)     
  McGovern (2010) GWAS#2b94814081.03(0.93-1.13)     
  McGovern (2010) GWAS#3b102225031.15(0.99-1.33)     
  Total616112541        

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Supporting Information

We comprehensively reviewed the published studies that examined the CD susceptibility loci in UC patients and performed a meta-analysis to clarify the common genetic factors for both diseases. We found associations at 25 out of 33 candidate loci. Among them, we confirmed the associations in 17 loci reported in the previous GWAS,20 and this study found an additional eight common susceptibility loci for CD and UC, namely, GCKR, ATG16L1, CDKAL1, ZNF365, LRRK2-MUC19, C13orf31, PTPN2, and SBNO2. Among these additionally identified loci, GCKR and LRRK2-MUC19 have never shown nominal association with UC in any single studies performed to date. Although the genetic risk of each locus was modest, many genes or loci will contribute to the pathogenesis of both CD and UC.

Previous GWAS identified that the autophagy-related genes are associated with the susceptibility of CD.6, 7, 9, 10, 13 In contrast to the strong association with CD, previous association studies for UC showed inconsistent results in these autophagy-related genes.16, 21–24, 29–31, 56 Our meta-analysis demonstrated nominal association with ATG16L1 by using 11,466 cases and 19,659 controls (P = 4.7 × 10−2, OR 1.05, 95% CI: 1.00–1.10). Other autophagy-related genes also showed associations with UC in this study (P = 1.54 × 10−2, OR 1.21, 95% CI: 1.03–1.41 for LRRK2-MUC19; P = 1.18 × 10−3, OR 1.14, 95% CI: 1.05–1.24 for IRGM). These findings suggest a possibility that autophagy might contribute to the development of both UC and CD, but its effect may be weaker for UC.

There is another possibility that the association of autophagy-related genes are caused by the contamination of colonic CD cases because rs2241880-rs10210302-rs3828309 (ATG16L1) and rs4958847 (IRGM) showed heterogeneity among the studies. However, we could not find any consistent set of studies that contributed to this heterogeneity. Moreover, when we assume the possibility of this misclassification for ATG16L1, colonic CD cases should be included in more than 20% of UC cases based on the assumption of a case–control study of 11,466 cases and 19,659 controls, an allele test model, a risk allele frequency of 0.571 based on the HapMap-CEU population, an allelic OR of colonic CD for 1.25,13 a statistical power of 0.80, and a P-value of 0.05. Since the diagnosis of UC was made by the established guidelines in each study, we think the association of autophagy-related genes in this study might not be caused by the misclassification of colonic CD cases in the previous studies.

Recent genetic studies have revealed shared genetic components of different immune-related diseases.66 For the shared susceptibility genes between CD and UC, previous studies have shown the importance of the common pathogenesis of the IL-23/Th17 signaling pathway, which promotes inflammation in the adaptive immune response.14 Many genetic variants including in this pathway such as IL23R, IL12B, JAK2, and STAT3 are associated with susceptibility for both diseases. Among the eight additionally identified common susceptibility loci for CD and UC, several genes are reported to be associated with various diseases or traits: C13orf31 is associated with leprosy.67PTPN2 is associated with type 1 diabetes68, 69 and celiac disease.70CDKAL1 is a susceptibility gene for type 2 diabetes.71–73GCKR is implicated in metabolic traits such as triglyceride,74–76 fasting glucose,77 and serum uric acid.78 However, there is little information how these genes affect the development of CD and UC. Functional analysis of these genes will provide further understanding of the common pathogenesis of CD and UC.

When we compared our results with those of a recent meta-analysis for UC,20 we could not find a significant association in the 6q21 locus. In the present study we performed a meta-analysis using the data of rs7746082 that showed the strongest association with CD at the 6q21 locus.13 However, the GWAS meta-analysis estimated the association using the data of rs6938089, best proxy SNP for rs7746082.20 Although the r2 value between rs7746082 and rs6938089 is 0.60 for the HapMap CEU population (release #27, build 36), there is a possibility that the hidden causative variant at the 6q21 locus might be different between CD and UC. Further detailed analysis is necessary to clarify the effect of the 6q21 locus on susceptibility to CD and UC.

Significant publication bias was observed at rs9292777 on 5p13 locus. The funnel plot showed that the largest study21 had the largest OR, whereas the OR of the smaller studies were all shifted to the smaller ones. Based on this asymmetrical distribution of OR, we excluded this SNP in this study.

In conclusion, in addition to the reported common susceptibility loci, we identified eight common susceptibility loci for CD and UC by a meta-analysis of published studies using more than 30,000 subjects. Our data indicate that UC and CD share many genetic factors with small effect. These findings will help to clarify the common pathway involved in the development of both diseases.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Supporting Information

We thank Atsushi Hirano for assistance with literature searches.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
IBD_21651_sm_SuppTable1.doc111KSupporting Information Table 1.
IBD_21651_sm_SuppTable2.doc428KSupporting Information Table 2.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.