Inflammatory bowel disease: Genetic and epidemiologic considerations

Abstract

Genome-wide association studies have firmly established that many genomic loci contribute to inflammatory bowel disease, especially in Crohn’s disease. These studies have newly-established the importance of the interleukin 23 and autophagy pathways in disease pathogenesis. Future challenges include: (1) the establishment of precisely causal alleles, (2) definition of altered functional outcomes of associated and causal alleles and (3) integration of genetic findings with environmental factors.

Key Words: Crohn's disease, Ulcerative colitis, Interleukin 23, Autophagy, Complex genetics

INTRODUCTION

The fields of genetics and epidemiology share a common goal of identifying specific factors, present in some individuals, and absent in others, that contribute to disease causation. As with many diseases, in inflammatory bowel disease (IBD), carefully performed epidemiologic studies provide an ongoing basis for the rationale and design of genetic studies. Observations on the nature and magnitude of familial clustering of IBD cases provided the framework for the present genetic observations that Crohn’s disease (CD) and ulcerative colitis (UC) share some susceptibility alleles (e.g. IL23R) whereas others are unique to one disease subtype or the other. This review summarizes present understanding of the genetics and epidemiology of IBD, and describes the encouraging early results from genome-wide association studies which demonstrate significant consistency between studies.

EPIDEMIOLOGIC OBSERVATIONS FORM THE BASIS FOR GENETIC STUDIES IN IBD

Familial risk in IBD

The observation that cases of IBD cluster within families suggested that both shared genetic and environmental factors could contribute to disease pathophysiology. That monozygotic (MZ) twin concordance is significantly higher than dizygotic (DZ) twin concordance for both CD (20%-50% vs 0%-7%, respectively) and UC (14%-19% vs 0%-5%, respectively) indicates that genetic factors definitively contribute to both disorders[1–4]. The observation that, even among MZ twins, disease concordance is significantly less than 100% demonstrates the significant role that environmental and developmental factors play in disease pathogenesis. Like most genetic disorders, inheritance in IBD does not follow a simple Mendelian pattern. This would indicate that no single gene mutation is sufficient and/or necessary to contribute to disease, indicating that IBD represents complex, multigenic genetic disorders. In complex genetic disorders, any single gene mutation/polymorphism will not confer a simple determinative effect of disease development (as is the case for Mendelian, single-gene disorders), but rather merely confers a statistically increased, but non-determinative susceptibility to developing disease. Approximately 5% to 20% of patients with IBD have a family history of IBD[56]. A family history of IBD is generally more prevalent among CD than UC patients[6]. The relative risk to siblings, λs, (sibling risk compared to general population risk) ranges from 30-40 for CD and 10-20 for UC[78]. Furthermore, the relative risk for UC to siblings of a CD proband was observed to be 3.9 and the cross-disease relative risk to siblings of a UC proband was observed to be 1.8, sugges-ting that CD and UC are genetically related[9].

Comparative prevalence of IBD

The prevalence of CD and UC is highest in North America, northern Europe, and the United Kingdom, with averages ranging from 100 to 200 cases per 100 000[10–13]. IBD is more common in European Americans compared with African Americans, and the lowest rates of IBD have been reported in Hispanics and Asians[714–17]. The prevalence of CD based on inpatient and outpatient visits in a Southern California Kaiser HMO study was 43.6 per 100 000 for European Americans, 29.8 per 100 000 for African-Americans, 5.6 per 100 000 for Asians and 4.1 per 100 000 for Hispanics[18]. However, two different Kaiser HMO studies found similar rates of hospitalizations among African-American and European ancestry individuals[1819]. In addition, a more recent study demonstrated a similar incidence of IBD in African-Americans and European Americans[20]. Within European ancestry populations, the rates of IBD are higher in persons of Jewish ancestry than other ethnic groups[1421], and higher in Ashkenazi (Central and Eastern European) Jews than in Sephardic (Middle Eastern and Spanish ancestry) Jews[1421].

The hygiene hypothesis

Generally, IBD incidence is lower in developing countries and higher in industrialized countries, such as in North America and Europe; as a country undergoes the transition to industrialization, the prevalence of UC rises first, followed by a rise in the prevalence of CD[10]. The hygiene hypothesis proposes that lack of exposure to select microbial agents in childhood causes IBD[22]. In support of this, in Manitoba, patients with CD are less likely to have lived on a farm and less likely to have drunk unpasteurized milk. Furthermore, CD patients were more likely to have used tap water rather than well water, and more likely to have had a pet[23]. Against this hypothesis, a separate study of pediatric IBD observed that owning a pet, day-care attendance, and “physician diagnosed infection” between 5 and 10 years of age was associated with increased CD risk, and regular use of a personal towel and lesser crowding in the home with decreased risk[24].

The measurable changes in disease incidence in various geographic regions highlights the contributory role of environmental factors in IBD pathogenesis. A leading culprit for such environmental contributions is a changing intestinal intraluminal microbial milieu. At birth, the intestinal lumen is largely sterile[25], with rapid bacterial colonization ensuing in the neonatal period. There is some evidence to suggest that fecal colonization patterns may reflect environmental cues in the perinatal period and thus may be manipulable[2627] as a potential means of preventing disease in high risk individuals.

Tobacco and IBD

Within Western countries, smoking is the most well established environmental risk factor, increasing risk for CD by approximately two-fold[28] and non-smoking decreasing risk for UC two- to five-fold[14]. Being an ex-smoker increases the risk of developing UC two-fold[29]. Ex-smokers make up an increasing percentage of older patients diagnosed with UC, accounting for more than 35% of the attributable risk of late onset (> 45 years) UC and a large component of the second peak in diagnosis[30]. Interestingly, Bridger et al observed that in IBD pedigrees with UC/CD sibling pairs, smokers tend to develop CD and non-smokers UC (O.R. 10.5)[31]. A significant increase in smoking in younger patients with familial CD has been reported[30]. Smoking is also associated with early recurrence of CD following surgery[32]. Future genetic studies in IBD will explore whether specific genetic associations will be stratifiable based on tobacco status at the time of diagnosis. However, tobacco may exert many of its phenotypic effects through epigenetic (reversible, changes in gene regulation that occur without a change in DNA sequence or genotype) mechanisms, such as DNA methylation effects on transcriptional activation[33].

HUMAN GENETIC VARIATION AND THE HAPMAP PROJECT

The most abundant of the human genetic variants are single nucleotide polymorphisms (SNPs). A SNP is a DNA sequence variation occurring when a single nucleotide, A, T, C, or G, in the genome differs between individuals, or between homologous chromosomes within an individual. Three salient features of SNPs include, (1) their allele frequencies within a population of interest, (2) their correlation, or linkage disequilibrium, with SNPs in the immediate genomic vicinity, and (3) whether or not the SNP of interest results in a measurable phenotypic change in protein function and/or expression, that is, represents a functional polymorphism.

As a very general rule, SNPs with a minor allele frequency approaching 50%, are genetically more ancient, having had a longer time to increase in frequency, and are more likely to be observed in all racial groups. In contrast, uncommon SNPs having a minor allele frequency of less than 5% (that is 5 of 100 chromosomes tested from 50 individuals carry the minor allele) are more likely to be evolutionarily more recent, and are often observed uniquely in one racial group or another. A number of factors besides the evolutionary age of a SNP can affect its frequency within a population, notably whether or not the SNP confers a selective advantage or disadvantage with respect to reproductive fitness within a population. It may be speculated that functional SNPs that contribute to the development of chronic inflammatory disorders such as IBD may confer a selective advantage with respect to combating historically significant infectious pathogens.

It is estimated that there are 5 million SNPs common SNPs having a minor allele frequency greater than 10%[34]. In order to comprehensively sample all common variation throughout the genome, however, it is not required that millions of SNPs be directly tested, due to the high degree of correlation, or linkage disequilibrium that exists between SNPs throughout the human genome. The Human HapMap Project empirically defined the linkage disequilibrium patterns in European, African and Asian cohorts and provided the basis for the genome-wide association studies that increasingly being reported for various complex genetic disorders[35]. It is estimated that genotyping several hundred thousand SNPs in European ancestry cohorts will sample nearly 80% of the common human variation with an r² (measure of linkage disequilibrium) of greater than 0.8[36].

The costs of genotyping several hundred thousand SNPs continue to decrease, and therefore genotyping sufficiently large cohorts to identify IBD genes through genome-wide association studies is now feasible. Once associations with SNPs are identified, the challenge remains to identify the functional polymorphisms that account for the statistical association. The precedent from Mendelian disorders would suggest that those amino acid variants which directly affect protein structure are likely to disproportionately contribute to disease susceptibility. Within amino acid polymorphisms, those variants in highly conserved (between species) regions or contained within key functional domains are more likely to have significant functional effects. Ultimately, however, proof of disease contribution requires demonstration of altered functional effects associated with the mutation of interest.

IBD GENETIC ASSOCIATIONS PRIOR TO THE ADVENT OF GENOME-WIDE ASSOCIATION STUDIES

Because the statistical effects observed in multigenic disorders are relatively modest, the hallmark for accepting genetic associations is replication in independent cohorts. However, interpretation of disease association studies has been complicated by the reporting of often conflicting studies. In such instances, meta-analyses of all reported studies can provide some insight, with the caveat that publication bias (e.g. negative studies may be more difficult to publish than positive ones) may occur.

Nod2 (CARD15) associations to ileal CD

The most well-replicated IBD genetic associations is the Nod2 gene association with ileal CD[3738]. Nod2 was the first definitive risk factor for CD and one of the first genes identified for a common complex genetic disorder. It functions as an intracellular sensor for bacterial peptidoglycan[39], and can be activated by a minimal bioactive component, muramyl dipeptide (MDP)[4041]. MDP is present in both gram positive and negative bacteria and activates NF-κB and MAP kinase pathways[42]. Three uncommon (minor allele frequency less than 5% in healthy controls) polymorphisms, Arg702Trp, Gly908Arg, and Leu1007fsinsC are each highly associated with CD. These findings have been extensively replicated by a number of subsequent studies. A meta-analysis of 39 studies showed an odds ratio for simple heterozygotes of 2.4 (confidence interval, C.I. 2.0-2.9), and for homozygous/compound heterozygous carriers of 17.1 (C.I. 10.7-27.2)[43]. Nod2 carriage has been specifically associated with ileal involvement, stricturing complications and a modestly earlier age of onset.

Each of the three CD polymorphisms are located within or near the leucine rich repeat, sensing domain of Nod2, and each results in a decreased capacity to activate NF-κB in response peptidoglycan or MDP stimulation[3944–48]. The frameshift mutation, Leu1007fsinsC demonstrates a largely complete deficiency in the capacity to signal in response to MDP stimulation[45]. Therefore, these functional polymorphisms represent direct risk alleles for CD. Importantly, the Nod2 discovery provides specific support for the long-held hypothesis that CD results from a genetically dysregulated host immune response to luminal bacteria. These Nod2 mutations have not been observed in Japanese, Chinese and Koreans with IBD[49–51], and they are rare in African-American IBD[52].

Despite the high odds ratios associated with the Nod2 mutations, it is estimated that the disease penetrance, even for homozygous or compound heterozygous carriers is limited, suggesting that these Nod2 variants alone are insufficient to produce disease. As each of the three major mutations demonstrates decreased function in primary human cells[3944–48], Nod2-deficient murine models provide an important model for human disease. The absence of intestinal inflammation in Nod2-deficient mice further highlights the insufficiency of this pathway alone to induce CD[4253]. Since the discovery of the CD-association with Nod2, a number of studies have added to understanding of the Nod2 functional pathway.

Given the dysregulation in intestinal immune homeostasis characteristic of CD, it is logical to hypothesize that Nod2 is important for either impaired tolerance mechanisms critical in limiting excessive activation of the intestinal immune system and/or altered initial defenses against intestinal bacteria. There are a broad range of tolerance mechanisms operating in the intestine to regulate excessive immune responses in the context of its unique exposure to a continuous bacterial load. Regarding initial defenses, a defect in bacterial recognition and responses at the intestinal epithelial surface might result in alterations in the population of commensal organisms, and/or increased invasion, which in turn, could contribute to an increased propensity toward intestinal inflammation. Nod2 is highly expressed in Paneth cells of the small intestine[5455] and Nod2 mutations are associated with ileal location of disease[56]. Studies in humans and mice have shown that normal Nod2 function is required for optimal defensin expression and therefore, impaired defensin expression and regulation may contribute to the increased CD susceptibility observed in Nod2 risk allele carriers[57–59].

IBD5 association on chromosome 5q31

Within the IBD5 linkage region, association of CD with a 250 kb region on chromosome 5q31 was reported[60]. This association has been subsequently replicated in a number of studies[61–64], for both CD and UC[6265]. Candidate polymorphisms found on this haplotype within the organic cation transporter OCTN1 (SLC22A4) and OCTN2 (SLC22A5) genes have been reported[6667]. However, statistically equivalent evidence for association has been observed throughout the risk haplotype[68] encompassing several genes, highlighting the need for additional genetic and functional investigation in this region. Phenotypic correlates have been reported for perianal disease[6169], colonic location[70], disease complications and progression[6871], extensive disease in UC[72], as well as for decreased height and weight at diagnosis in pediatric cohorts[73]. In contrast, some studies have not observed clear genotype-phenotype correlations[7475].

HLA associations

Replicated HLA class II associations in IBD include HLA-DRB1*1502 (serological marker HLA-DR2)[76] association with UC, and HLA-DRB1*0103 association with UC and colonic CD[7778]. HLA-DRB1*0103 is noteworthy in that it is a risk factor for both UC and colonic CD, suggesting it may play a role in chronic inflammation of the colon independent of major IBD phenotype (i.e., CD or UC)[78]. As with IBD5, the DRB1*0103 and DRB1*1502 class II variants are in strong linkage disequilibrium with SNPs on multiple immunologically active candidate genes. Present approaches have not been able to discern whether these other genes or the class II genes are the true risk genes in the HLA locus.

Reported, but unconfirmed IBD associations

Associations which have been reported within linkage regions to IBD but have not been consistently replicated include an indel polymorphism in the promoter region of the NF-κB1 gene on chromosome 4q[79–82], multiple polymorphisms in the MDR1/ABCB1 gene on chromosome 7q[83–92], and polymorphisms in the DLG5 gene on chromosome 10q[7093–102]. (See section C.4.1.) Haplotypes in the terminal exons of the chromosome 7p Nod1 (CARD4) gene and a Nod1 (CARD4) indel polymorphism were associated in two different study populations[47], although other studies have not replicated evidence for association[103–105]. Finally, studies of candidate genes that are not located within IBD linkage regions have revealed weak associations between UC and a variable number of tandem repeats polymorphism in the IL1RN gene[106] and between IBD and the TLR4 gene[107–113]. In a large case-control study, association was observed in UC and CD for Ala1011Ser within the Myo IXb (MYO9B) gene[114] that had been previously been implicated in celiac disease[115]. If replicated, this would indicate shared susceptibility pathways between multiple types of intestinal inflammation.

GENOME-WIDE ASSOCIATION STUDIES IN IBD

Emerging technologies now provide for the direct testing of a large number of SNPs throughout the genome in genome-wide association studies. The multiple genome-wide association studies reported below vary with respect to the genotyping platform utilized and the population cohort tested. Despite these differences, however, the first glimpses of comparative results between the genome-wide association studies are demonstrating encouraging consistency of findings between studies. Taken together, the early results from these studies would suggest that significant advances in defining well-replicated gene associations in IBD, especially in European ancestry cohorts, will shortly ensue.

TNFSF15 association to CD

The first genome-wide association study in IBD involved testing nearly 80 000 SNPs in a Japanese CD cohort[116]. The most significant findings implicated SNPs and haplotypes within the TNFSF15 (TNF superfamily) gene. These results were subsequently confirmed in European IBD cohorts[117]. TNFSF15 is a Th-1 polarizing cytokine which has been reported to be increased in IBD mucosa[118]. In a separate Belgian CD cohort, however, no significant evidence for association was observed[119]. These differences could reflect differences in markers genotyped and/or different contributions of susceptibility genes between Asian and European IBD cohorts. Future studies in a variety of population cohorts will provide important comparative insight.

IL23R is associated with CD and UC

A genome-wide association study testing over 300 000 autosomal SNPs was recently reported[120]. Three SNPs had nearly two orders of magnitude greater significance compared to the next most significant markers. Two of the three markers, were in the known CD susceptibility gene, Nod2. The third marker, rs11209026 (P = 5.05 × 10^-9, corrected P = 1.56 × 10^-3), was a non-synonymous SNP (Arg381Gln) in the IL23R gene on chromosome 1p31. Replication of these findings was observed in an independent cohort of 883 nuclear families and observed significant association in CD and non-Jewish UC. Therefore, IL23R represents both a CD and a UC susceptibility gene. The less common glutamine allele of Arg381Gln has an allele frequency of 1.9% in non-Jewish CD and 7.0% in non-Jewish controls and therefore protects against IBD[120]. Other variants within IL23R are also associated with IBD, independent of the Arg381Gln association. Since the initial report, these findings have subsequently been replicated in Scottish pediatric IBD[121] and Belgian CD[119] cohorts, indicating a clear role for IL23R in IBD susceptibility.

Given the IL-23 pathway’s role in activation and perpetuation of organ-specific inflammatory responses, the genetic findings suggest that targeting the IL-23 pathway may be a rational therapeutic approach. In this regard, anti-p40 administration, which blocks both IL-23 and IL-12 activities, has proved promising[122]. The contribution of the IL23R pathway to IBD will likely involve more than simple gain or loss of function IL23R variants and ongoing studies of this pathway may reveal new therapeutic options. Future studies should examine mechanisms of the strong protective effect of the Arg381Gln allele could potentially be exploited to define clinical outcomes. The genetic association and the pro-inflammatory role of IL-23 strongly prioritize this pathway as a therapeutic target in IBD.

The IL23R genetic association to IBD correlates precisely with recent immunologic advances in understanding of the IL-23 pathway. A number of murine models of colitis, including IL-10 deficiency, T cell mediated (CD45Rbhigh reconstituted in RAG deficiency) and non-T cell mediated (agonistic anti-CD40 RAG deficiency, H. hepaticus-induced) colitis demonstrated significant amelioration when crossed with p40 and p19-deficient mice. This demonstrates the requirement for an intact IL-23 pathway for a variety of intestinal inflammation pathways[123–126].

A complete understanding of the strongly protective effect of IL23R polymorphisms in IBD should not be solely restricted to pharmacologic approaches to mimic the Arg381Gln polymorphism. While it is logical to test anti-p19 antibodies in IBD[127], the underlying paradigm is one of an ongoing requirement for continued immune suppression. However, it is unlikely that the approximately 14% of European ancestry individuals heterozygous for the glutamine allele harbor significant immunodeficiencies, and therefore these individuals provide a unique opportunity to better understand the lifelong, dynamic host-intestinal microbial interactions that are the key to IBD. We are born with a sterile intestine[25] and an immature immune system[128], with the rapid acquisition of intestinal flora evolving dynamically with early instruction signals to the host immune response. The significant susceptibility of neonates to infection may correlate with their incapacity to induce IL-12p35 with lipopolysaccharide stimulation[129], with IL-23p19 induction being largely intact[130]. This would suggest that functional IL-23 pathway polymorphisms may be particularly important in modulating neonatal development of intestinal tolerance and bacterial colonization.

ATG16L association to CD

In a genome-wide survey of nearly 20 000 nonsynonymous SNPs, an amino acid polymorphisms, Thr300Ala within the ATG16L1 gene was found to be highly associated with CD. The ATG16L1 protein is comprised of N-terminal APG16 domain consisting of coiled coils and eight C-terminal WD repeats. The Thr300Ala variant is located at the N-terminus of the WD-repeat domain in ATG16L1. The ATG16L1 gene is part of the autophagosome pathway and has been implicated in the processing of intracellular bacteria. Of interest, no association was observed in UC and a statistical interaction was reported with the Nod2 (CARD15) CD associations. Since the initial report, this association has been confirmed in a separate, Belgian CD cohort[119]. In addition, in this cohort, no association was observed in UC. Taken together, the ATG16L1 association represents a well-replicated CD-specific association.

Association of a gene desert on chromosome 5p13.1 which modulates the expression of the prostaglandin receptor EP4 (PTGER4)

In a genome-wide association study in a Belgian CD cohort, in addition to the CARD15, IL23R and ATG16L1 associations, association was observed in a region on chromosome 5p13.1. The most significant association was observed in a gene desert region flanked by a number of potential candidate genes, including CARD6, complement factors C6, C7 and C9 and the prostaglandin receptor, EP4 (PTGER4). No association was observed in UC for markers in this region. PTGER4 is a compelling candidate, in part because PTGER4 deficient mice develop a more severe colitis with dextran sodium sulfate treatment[131]. In elegant analyses, the investigators compared SNP data in this genomic region with mRNA expression levels of flanking genes from the corresponding, individual lymphoblastoid cell lines. Throughout the region of association, a number of SNPs were significantly associated with mRNA expression levels of PTGER4, including at least one SNP demonstrating both CD association as well as correlation with PTGER4 expression. However, the susceptibility allele at this marker corresponds with increased PTGER4 expression, which is not consistent with the increased colitic susceptibility observed in PTGER4-deficient mice[131]. However, taken together, these findings strongly suggest that genetic variation in this region is associated with IBD and significantly regulates PTGER4 expression.

Assessment of genome-wide association studies in IBD

It is anticipated that the most significant associations observed in various genome-wide association studies will be largely replicable between studies in comparable population cohorts. Genome-wide association studies provide a comprehensive, unbiased landscape of common variation contributing to disease. Three relative limitations to these studies should be noted, however. Apart from the highly significant, outlying associations such as those observed for the Nod2 and IL23R gene associations, many of the most biologically significant disease associations may confer less significant statistical associations, and be obscured by surrounding “noise” conferred by false positive associations. Study design approaches to dissect true disease associations from surrounding noise remains a major methodological challenge moving forward. A second limitation is that the genotyping platforms typically utilized for these studies are limited to testing common genetic variation, with minor allele frequencies of greater than 5%. In the presence of significant allelic heterogeneity for uncommon variants within a disease gene, genome-wide association approaches will be relatively insensitive. The Nod2 associations are observed in genome-wide surveys so clearly because all three of the uncommon CD variants by chance share a common haplotype background. Therefore, the identification and characterization of uncommon variants may largely be missed by a complete reliance of genome-wide association approaches. Finally, many of the early genome-wide genotyping platforms were specifically developed to efficiently test European ancestry cohorts, and may not as efficiently assay Asian or African populations. Particularly for African populations which have shorter regions of linkage disequilibrium, testing of larger numbers of genetic markers will be required.

From genetics to genetic epidemiology: defining risks in patient subsets to assist clinical practice

A more complete pathophysiologic genetic definition of IBD will require integration of clinical observations, genetic data, statistical analyses, and delineation of underlying biologic processes. Interrogating GWA data for gene-gene interactions analyses has been proposed as being more powerful than traditional, single locus analyses under a range of interactive models. However such analyses may be more powerful if prior biologic knowledge on signaling pathways, in vivo models of inflammation and integration of genetic insight is applied. The growing catalogue of functional polymorphisms contributing to distinct chronic inflammatory disorders, while not necessarily demonstrating disease association, may contribute to IBD pathophysiology as disease modifiers. Stratification of genetic associations by phenotype variation and/or use of covariates may benefit genetic understanding of IBD by reducing heterogeneity that can obscure association evidence. Because of the strength of the Nod2 association in CD, and because ileal CD represents a large fraction of overall cases of CD, the Nod2 associations were easily observed in the “all CD” analyses, and even the “all IBD” analyses. However, as the Nod2 genotype risk represents one of the largest among complex disorders, it would be anticipated that many IBD association signals may require correct subsetting, of which disease location is most well-established. Integration of clinical and genotype data may provide the capacity to predict complications and disease course in a clinically useful way. While the integration of definitive IBD risk alleles would be the most straightforward factors to include in such predictive models, it is possible that many key genetic factors (particularly well-established functional polymorphisms from related disorders) may, by themselves, not be associated with the major phenotypes, but rather function solely as disease modifiers. For example, it is possible that certain key genetic factors (particularly well-established functional polymorphisms from related disorders) may, by themselves, not be associated with the major phenotypes but rather serve to modify disease expression. Detecting such relationships requires a conditional analysis restricted to those with IBD in which certain phenotypic characteristics are modeled as a function of the presence of one or more genetic variants. Such models will also provide a framework for generating predictions about clinical course among newly diagnosed cases-providing clinicians important information for use in treatment planning.

The identification of major single gene associations such as IL23R and Nod2 provides a framework around which risk models for multigenic disorders can be developed. A major research interest moving forward will be identifying gene-gene interactions that contribute to increased disease risk or define pathophysiological subsets within the disease. This will be explored through: (1) a broad-based, systematic search of genomic regions demonstrating the most significant evidence for association, (2) pathway analyses integrating specific knowledge regarding functional human polymorphisms relevant to the IL23R and Nod2 pathways, and (3) studying key mechanistic intermediates known to be particularly relevant to disease pathogenesis.

As functional genetic variants that predispose to IBD are identified, their effect on modifying disease phenotype and/or disease course will need to be examined. For example, the CARD15 mutations increase susceptibility to CD generally, affect disease location (increase susceptibility to ileal location), and modify disease behavior. Because of the challenges of longitudinal follow-up, prospective evaluation of IBD-associated mutations represents a major challenge. The efficient abstraction of critical research datapoints from active clinical practices in an accurate and reproducible manner will be required. Interpretation of disease course from multi-center studies given differences in practice patterns represents an additional interpretative challenge. However, such studies will provide the basis for improved translation of genetic discovery in IBD and will need to be designed via active communication and collaboration throughout the IBD research community.