• Inflammatory Bowel Disease
• Genetic Testing
• Genetics

The paper by Tyler et al1 describing genetic associations in patients with ulcerative colitis (UC) preoperatively who then develop chronic pouchitis (CP) and a Crohn's disease (CD) like phenotype postoperatively further extend our clinical definitions of CD and UC based on an evolving knowledge of inflammatory bowel disease (IBD) genetics. Many described human IBD susceptibility genes are shared between UC and CD, highlighting overlapping clinical and pathophysiological features. However, single nucleotide polymorphisms (SNPs) in NOD2 (nucleotide-binding oligomerisation domain 2) are the most important genetic risk factors for the occurrence of CD. In individuals of European ancestry, heterozygous carriage of one of the three major risk alleles confers a 2.4-fold increase in risk for CD; homozygous or compound heterozygous carriage confers a 17.1-fold increase in risk for CD.2 ,3 Clinically, NOD2 mutations have been consistently associated with ileal location and stricturing disease.3 ,4 Of the major risk alleles, the insertion mutation NOD2insC which codes for a truncated protein, confers the greatest risk for CD.2

NOD2 is a member of the NOD-Like Receptor (NLR) (NOD, leucine-rich repeat-containing protein) family of intracellular sensors of microbes and microbial products.5 Progress has been made in understanding NOD2 function in relation to CD pathogenesis. NOD2 has been implicated in the expression of antimicrobial peptides which consequently shape the composition of the enteric microbiota.6 Macrophage and dendritic cell NOD2 regulates the expression of proinflammatory and anti-inflammatory cytokines central to IBD pathogenesis.7 Emerging evidence in mice and humans suggests that NOD2 represents a global regulatory pathway for small bowel inflammation. In mice, loss of Nod2 function alone does not cause intestinal inflammation. However, Nod2-deficient mice express low levels of antimicrobial peptides and reduced capacity to clear enteric microbes.8 Moreover, Nod2-deficient mice in which granulocyte macrophage colony stimulating factor (GM-CSF) was neutralised (to recapitulate the effect of endogenous anti-GM-CSF antibodies detected in CD patients) demonstrate mucosal barrier defects and develop transmural ileitis following exposure to a non-steroidal anti-inflammatory drug.9 These experimental findings indicate that NOD2 mutations may facilitate invasion and/or impair clearance of enteric bacteria, contributing to inflammation in ileal CD. In humans, evidence links NOD2 to other forms of small bowel inflammation. In graft versus host disease following allogeneic stem cell transplantation, NOD2 mutations in recipients and donors correlated with transplant-related mortality and severe gastrointestinal graft versus host disease.10 ,11 Moreover, NOD2 SNPs have been associated with combined liver and intestinal failure in subjects with short bowel syndrome as well as allograft failure following transplantation.12 ,13

The study by Tyler et al1 presents the largest published analysis of genetic risks for pouch-related complications following total proctocolectomy and J pouch surgery for UC. Pouch-related complications were defined as ‘Cronic Pouchitis’ (CP) and a ‘CD like’ (CDL) phenotype. The strongest single risk for CP and CDL was the presence of the NOD2insC polymorphism. Comparable with previous analyses of non-IBD controls in Caucasian populations, this SNP was present in 1% of the population that did not develop chronic pouch inflammatory complications. NOD2insC was found in 5% of patients with CP and 6% of patients with the CDL phenotype. Hence, based on allele frequency in each clinical group, the majority of patients who possess NOD2insC will develop CP or CDL phenotype, and in fact this analysis likely underestimates the problem as it is plausible that patients in the non-CP group who possess the NOD2insC allele eventually develop CP or CDL features. However, the vast majority of patients who developed CP or CDL presentation did not have NOD2insC. Accordingly, the investigators next created a multivariate risk model for the development of CP and the CDL phenotype. A model containing three genetic polymorphisms including NOD2insC combined with a serological marker (anti-CBir1) and the clinical risk factor of smoking performed significantly better in predicting pouch-related inflammatory complications than models that did not include genetic markers (sensitivity 80.0%, specificity 70.3%).

Two aspects of this important study should be critically addressed. The first point relates to research focusing on the ileal pouch and pouchitis. Approximately one fifth of UC patients may undergo surgery. The surgical treatment of choice in UC is restorative proctocolectomy, with the creation of an ileal J pouch. While pouch surgery is generally safe, offers good quality of life and has a satisfactory long-term functional outcome,14 it is associated with a substantial rate of complications. Pouch inflammation, pouchitis, is the most prevalent long-term complication in UC patients, with a reported incidence of up to 60%.15 Thus, understanding the reasons for pouchitis development and severity is meaningful for these patients. As the vast majority of patients undergoing pouch surgery are UC patients who by definition had a disease restricted to the colon, as was the case in this study, the de novo development of inflammation in the small bowel reservoir comprising the pouch, or the small bowel loop proximal to the pouch (the afferent loop) would be unexpected if our clinically based, ‘old world’ definitions were correct. The authors of this study assessed clinical, serological and genetic factors predicting the development of pouchitis. They show that the presence of genetic risk for CD correlates with pouch-related complications. This data support the notion that clinical definitions of UC and CD represent a continuum that ultimately may be better defined by genetics and immune responses than by rigid clinical criteria established a century ago. The basis for the genetic analysis is stringent clinical phenotyping of the patient population. Here, the authors chose to divide patients into three major groups. A ‘no pouchitis’ group included patients without an episode of pouchitis, as well as those who experienced no more than four acute pouchitis episodes per year, each responding to a 2 weeks course of antibiotics. This group was compared with the ones with a more aggressive disease phenotype, the CP and CDL groups. While the CP/CDL groups can possibly be grouped together for analysis purposes, representing the most aggressive forms of pouch inflammation, the authors decision to group patients with a normal pouch, usually defined as those not experiencing symptoms of pouchitis and not being treated, with those experiencing up to four episodes of pouchitis/year is more problematic. The latter group comprising recurrent acute pouchitis patients, especially those having a flare every 3 months, certainly present a distinct phenotype that cannot be easily distinguished from those with CP. Thus, separating the strictly normal pouch patients from the ones with CP/CDL, even at the cost of excluding patients with recurrent acute pouchitis from the analysis, could have added more meaningful knowledge. We suggest that, IBD, including pouchitis, is a spectrum of diseases that should be defined based on genetic, microbiological and immunological features, rather than clinical ones. However, to accomplish this goal in the near future, meticulous distinction between known phenotypes should be performed.

The second question to be addressed is a fundamental one: is the risk model suggested by the authors ready for integration into clinical practice? This question becomes more relevant and controversial as in the USA, NOD2 genetic testing is commercially available. Although this is a large analysis, there were still only 118 patients with CP and 109 patients with a CDL phenotype on which to base conclusions. The vast majority of patients who developed pouch complications did not have NOD2 polymorphisms, and even the multivariate risk model with a sensitivity of 80% is imprecise for making major clinical decisions such as initiation of preventative postoperative medications or avoidance of surgery altogether in the absence of any data suggesting that outcome can be altered. In fact, the potential misuses of applying this model in clinical practice could have unfavourable clinical consequences based on our existing state of knowledge. The most egregious misuse of the proposed model would be to counsel patients against medically indicated J pouch surgery. Likewise, after J pouch surgery, it is impossible to recommend any preventative medical regimen (akin to postoperative CD) based on genetic, serological and clinical risk stratification in the complete absence of data suggesting that postoperative medical therapy will alter the development or long-term outcome of CP or a CDL phenotype. Dissecting the components of the multivariate model, the most important intervention one could advocate would be smoking cessation in UC patients that continue to smoke after surgery, which is not a new concept. In light of these comments, is this information in any way applicable to clinicians and patients? Speculatively, if patients are defined as high risk for pouch-related inflammatory complications, then an endoscopic surveillance and treatment algorithm similar to our clinical practice in postoperative CD can be contemplated. However, before blindly adopting such a strategy, more data are necessary. First of all, the distinction between CP managed with antibiotics and CP refractory to antibiotics and CDL inflammation requiring immunomodulation, tumor necrosis factor inhibitors and repeated surgeries is clinically important, and cannot be discerned by this model. Based on this study, difficult but necessary multicentre analyses to prospectively test and refine the postoperative risk model will ultimately help guide clinical practice. Moving forward, perhaps this risk model can be strengthened by incorporation of longitudinal changes in the enteric microbiota following pouch surgery, as there is likely to be important relationships between genetic background and microbial composition which influence the development of pouch-related inflammatory complications.

In summary, this study sheds further light on the functional role of NOD2 in intestinal inflammation. Although initially described as a CD susceptibility gene, NOD2 is now established as a risk factor for pouch-related inflammatory events in UC patients, further illustrating the clinical continuum within the human IBDs. Although this new genetic information should not appreciably alter clinical practice at present, this study is an important first step towards establishing a clinically applicable risk model for postoperative complications after J pouch surgery which hopefully can eventually lead to better outcomes.