Personalizing Therapy for Inflammatory Bowel Diseases

Ashwin N Ananthakrishnan


Expert Rev Gastroenterol Hepatol. 2013;7(6):549-558. 

In This Article

Can the Natural History of IBD Be Predicted?

Both CD and UC are characterized by heterogeneity in their natural histories. At the time of diagnosis, CD tends to be predominantly inflammatory with increasing likelihood of penetrating or stricturing complications over time.[3,27] In a French cohort, the proportion of patients with stricturing or penetrating complications 20 years after diagnosis was 18 and 70%, respectively.[27] Natural history of postoperative CD suggests a similar heterogeneity in disease behavior with a non-uniform distribution of disease recurrence. In a landmark study by Rutgeerts et al., 39% of patients after 'curative' ileocolonic resection for their CD demonstrated no or minor endoscopic recurrence and had low rates of clinical symptoms while 44% of patients demonstrated early aggressive recurrence, were nearly universally symptomatic and had a high likelihood of repeat resection.[28] In UC, only a fifth of patients eventually require colectomy and a substantial proportion have a mild relapsing–remitting course.[29]

Clinical Predictors of Aggressive Disease

Prediction of aggressive disease behavior has remained challenging. In an elegant study, Beaugerie et al. examined the natural history of 1123 CD patients with at least 5 years of follow-up to identify clinical predictors early on in the natural history that would predict subsequent disabling disease.[30] The authors defined disabling disease as requiring hospitalization or surgery for CD, multiple courses of steroids, immunosuppressive treatment or persistently disabling symptoms interfering with functioning. While this overtly broad definition of disabling disease resulted in 85% of their cohort being classified as having disabling disease, nevertheless useful clinical clues emerged. Patients who were younger than 40 years at diagnosis (OR: 2.1), required steroids at diagnosis (OR: 3.1), or had perianal disease (OR: 1.6), were more likely to meet the definition of disability.[30] Loly et al. similarly attempted to predict disabling CD using a more restrictive definition of severe disease in a cohort of 361 patients.[31] In their study, disabling CD was defined as complex perianal CD, colon resection, two or more small bowel resections or the need for a permanent stoma and occurred in 58% in their cohort. Similar to the findings of the study by Beaugerie et al., perianal disease and need for steroids (but not age at diagnosis) were predictive of severe disease. However, additional predictors of severity identified in this cohort were weight loss of greater than 5% at presentation and stricturing behavior.[31]

In an examination of the Olmsted county cohort by Thia et al., ileal or ileocolonic location and perianal disease were associated with development of internal penetrating disease,[32] consistent with the analysis by Cosnes et al. associating ileal involvement with penetrating complications.[27] In addition, upper gastrointestinal involvement, male gender, current smoking, early corticosteroid use and penetrating disease were associated with a higher cumulative probability of major abdominal surgery.[32] Indeed, current smoking is the sole environmental factor that has been consistently associated with both a short- and long-term impact on CD.[33,34] In intervention studies, cessation of smoking is associated with a reduction in disease activity in the subsequent year while continued smoking in epidemiologic studies is associated with increased risk of initial and subsequent surgeries, and occurrence of disease complications.[33–35]

Another clinical scenario where prediction of disease course is important given heterogeneity in disease course is in the postoperative setting after an ileocolonic resection for CD.[36–38] Similar to the preoperative setting, current smoking was associated with higher rates of endoscopic and clinical recurrence and need for repeat surgical intervention.[39,40] Patients who undergo initial surgery for penetrating disease are more likely to require a repeat resection as are those with ileal disease.[41,42] Other clinical risk factors inconsistently supported by the literature include male gender, length of resected segment, presence of granulomas, histologic activity in the resected margin and type of anastomosis.[40]

There have been fewer studies attempting to predict natural history of UC. Extent of disease is an important predictor of requiring colectomy. Pancolitis at diagnosis conferred a threefold increase in risk of colectomy at 10 years in the IBSEN cohort.[29] In the Manitoba IBD cohort, male sex and requiring hospitalization at diagnosis were associated with increased risk of colectomy.[25] Both cohorts showed that older age was associated with reduced risk of colectomy. However, in an earlier analysis of the IBSEN cohort, disease extent was not associated with disease activity or course of disease over the first 5 years after diagnosis.[43]

Serologic Predictors of Aggressive Disease

The key mechanism underlying the development of IBD is a dysregulated immune response to commensal microbiota in a genetically susceptible host.[44] Consequently, a number of antimicrobial antibodies have been identified in patients with IBD, particularly CD.[45–48] Considerable interest lies in whether such biomarkers can be used to predict the natural history. Many of the studies examining this question have been retrospective and sometimes measured the levels of antibodies after the event of interest (penetrating complications or surgery) has occurred.[49,50] A higher overall titer or quartile sum of the antimicrobial antibodies is associated with ileal disease location, development of penetrating or stricturing complications and need for intestinal surgery.[49] Among the individual biomarkers, antibodies to Saccharomyces cerevisiae (ASCA), outer membrane protein (OmpC) and flagellin (anti-CBir1) have also independently demonstrated association with complicated disease.[51] In two prospective Canadian pediatric CD cohorts by Amre et al., antimicrobial positivity at baseline was associated with relapsing disease course, and an earlier time to occurrence of complications but not surgery.[52,53]

The repertoire of antibodies has recently expanded to include antiglycan antibodies against carbohydrate epitopes. Such antibodies also remain stable over the course of disease and are associated similarly with occurrence of complicated disease.[46,54] In a prospective cohort of 149 CD patients by Rieder et al., positivity for ASCA, anti-mannobioside IgG (AMCA), anti-chitobioside IgG (ACCA) and anti-laminarin IgA (anti-L) were each independently and cumulatively associated with shorter time to disease complications (fistulae or strictures) and need for surgery.[55]

Less data exist regarding the utility of such markers in UC. In a small study from the Mayo clinic by Sandborn et al., pANCA positivity was more common in treatment-resistant UC patients compared with those responding to treatment.[56] However, pANCA positivity wasn't associated with need for colectomy. Several groups have examined the role of such antibodies in predicting the likelihood of pouch complications in UC patients. In a study of 95 patients who underwent a total colectomy and an IPAA for management of their refractory UC, 63% were positive for pANCA. Stratifying by pANCA status, chronic pouchitis occurred in 56% of patients with high pANCA positivity compared with only 20% of those who were pANCA negative.[57] Other cohorts had identified associations between ASCA positivity and postoperative pouch fistulae,[58] CBir1 positivity and acute pouchitis[59] and a combination of ASCA and CBir1 positivity with chronic pouchitis and Crohn's-like disease of the pouch.[60]

Genetic Predictors of Aggressive Disease

An individual's genotype remains unchanged throughout life and could be used as a stable predictor of natural history of disease. While recent genome-wide association studies and meta-analyses have expanded the spectrum of risk loci,[61] much of the data associating genotype to phenotype have been for the NOD2 locus. A meta-analysis by Adler et al. summarized 49 relevant studies comprising 8893 patients, among whom 2897 had NOD2 mutations.[62] The pooled relative risk for complicated disease (defined as the presence of strictures or fistulae) for the presence of at least one NOD2 variant was 1.17 (95% CI: 1.10–1.24). The effect was primarily driven by homozygosity at the locus with a non-significant association for heterozygotes. Among the three common NOD2 variants, G908R was the only one meeting statistical significance for complicated disease and with individual disease phenotype separately. Not surprisingly, given the strong association between ileal location, penetrating or stricturing disease behavior and surgery, the presence of one or more variant NOD2 alleles was associated with a higher risk of surgery.[50,62–65] Genetic variants in the autophagy and adaptive immunity pathway may also confer risk for complicated disease. The G allele at the IRGM locus conferred a nearly 10-fold increase in risk for penetrating disease[66] while a variant in the IL-12B gene (rs1363670) increased risk of stricturing disease.[67,68] In an independent study by Sehgal et al., the IRGM locus was associated with increased risk for surgery with a shorter interval between resections.[69]

In addition to their independent effects, genetic variants may exert influence cumulatively. In a large Dutch study, Weersma et al. studied 1684 patients with CD and 1120 with UC who were genotyped for five risk loci: NOD2, IBD5, DLG5, ATG16L1 and IL-23R. Increasing genetic burden was associated with a more severe disease course that included early age of onset, penetrating or stricturing complications and need for intestinal surgery.[70] There are far less data examining influence of genetics on disease phenotype in UC; polymorphisms in the HLA-DRB1 locus and multi-drug resistance gene (MDR1) have been associated with extensive UC, while TNFSF15, IL-10 and IL-12B polymorphisms have been associated with need for colectomy.[71–75]

While the effect of individual genetic variants may be weak and replication data are lacking, there is some evidence supporting the overall influence of genetics on disease phenotype, offering promise in risk modeling. In addition, the influence of genetics may be additive to serologic biomarkers such as antimicrobial antibodies[50] or environmental variables through gene–environment interactions,[68] an area where further study is needed before translation into regular clinical practice.