Predicting Complicated Crohn's Disease and Surgery

Phenotypes, Genetics, Serology and Psychological Characteristics of a Population-Based Cohort

J. D. Ryan; M. S. Silverberg; W. Xu; L. A. Graff; L. E. Targownik; J. R. Walker; R. Carr; I. Clara; N. Miller; L. Rogala; C. N. Bernstein


Aliment Pharmacol Ther. 2013;38(3):274-283. 

In This Article


Cohort Description

The Manitoba IBD Cohort Study is a prospective, longitudinal study that aims to identify determinants of IBD outcomes. This population-based sample of adults was recruited within 7 years (median 4 years) of their diagnosis of IBD. They have been tracked prospectively through semi-annual surveys using standardised and validated self-report instruments, and annual clinical interviews and blood draws were completed by research nurses. The participants were originally recruited from a validated population-based research registry that has been described elsewhere.[31–34] Briefly, all adults with IBD in the province are included in a population-based database (the University of Manitoba IBD Epidemiology Database), drawn from the comprehensive provincial health administrative database using a validated definition of IBD. All those in the University of Manitoba IBD Epidemiology Database were invited to participate in a research registry. The University of Manitoba IBD Research Registry has subsequently been updated periodically with individuals identified from the administrative health database. Just over half of all the provincial cases of IBD (n = 3192) were enrolled in the Research Registry at the initiation of this study.[31]

Of those enrolled in the Registry, 606 were eligible for this study, based on the criteria of adult age and recent disease onset. The final Cohort was comprised of 388 who completed the baseline survey and interview, after those who declined (14%) or could not be reached (17%) were excluded. The Cohort has been established as having excellent representativeness of the provincial IBD population, with comparative age distribution, sex distribution and rural/urban residence.[35] Of the final, Cohort 182 had CD. Of these 182, 127 had paired serum from both time points of this substudy. In comparing the 127 participants in this study with the 182 population-representative participants that are in the overall Cohort Study, this study sample had comparative age and gender distribution, rates of smoking, family history of IBD and marital status distribution. Their CD behaviour and location, and surgery frequency are also similar.

Although we have used the Cohort Study to explore a variety of psychological issues in relation to IBD, it provides a unique population-based opportunity to explore whether psychological variables as well as serological and genotypic variables are associated with specific CD phenotypes and the need for surgery. Cohort participants were prospectively followed up with semi-annual surveys assessing disease activity and psychological functioning. A detailed description of the study population has previously been published.[32] As our cohort was drawn from the general population, there was significant variability in the diagnostic investigations undertaken before inclusion. For this study, baseline CD phenotype was determined using the Montreal classification within 3 years from diagnosis (T1) and repeated 5 years after enrolment in the Cohort Study, which equated to a median of 9.3 years from diagnosis (T2). The 3 years from diagnosis allowed time for adequate diagnostic investigations to occur to effectively classify a baseline phenotype. Blood work for serological and genetic analysis was obtained at enrolment in the Cohort Study and at T2.

Assessment of Disease Characteristics

We assessed three main outcomes at T1 and T2. These outcomes were as follows: (i) disease location as defined by the Montreal classification;[2] (ii) complicated disease, defined as stricturing (B2) or penetrating (B3) behaviour; and (iii) abdominal surgery for CD, excluding perianal surgery. Other characteristics recorded at baseline included, extra-intestinal manifestations of CD, family history of IBD and smoking status. Personality characteristics were assessed using the Neuroticism, Extroversion, Openness (NEO) Five-Factor Inventory (NEO-FFI).[36] The NEO-FFI is a well-validated 60-item scale designed to give quick and reliable measures of the five major domains of adult personality. The domains include neuroticism, extraversion, openness to experience, agreeableness and conscientiousness. Psychiatric diagnoses and adverse childhood experiences including sexual and physical abuse were determined through a semi-structured clinical interview, the Comprehensive International Diagnostic Interview (CIDI), which identified lifetime prevalence of anxiety, mood disorders and phobias.[37,38] Self-description of usual medication adherence was assessed with a validated self-report measure, the Medication Adherence Report Scale.[39,40]

DNA samples were obtained from peripheral whole blood samples collected from 167 subjects at T1. Samples were genotyped using the Illumina Golden Gate custom SNP assay on Illumina BeadStation500G. Genetic analysis focused on SNPs rs2066845, rs2076756 and rs2066847 in NOD2, rs3828309 and rs2241880 in ATG16L1, and rs11465804 in IL23R, which had previously been determined to be associated with CD in a separate population-based sample of Manitobans.[41] There was insufficient sample for adequate DNA isolation for 15 participants, and 40 had insufficient serum either at baseline or at 5 years, such that paired serum was not available. This left 127 participants with complete data to be assessed including paired sera at T1 and T2. All assays were performed at Prometheus Laboratories using a commercial assay (IBD-Serology 7, Prometheus Laboratories, San Diego, CA, USA) that included pANCA, ASCA IgA and IgG, anti-I2, anti-OmpC and anti-CBir1. Serology results at T1 and T2 were then compared.

Statistical Analysis

Descriptive statistics of demographic variables were generated using SAS version 9.2. WilcoxonRank–Sum Test and Chi-Square testing was used to compare phenotype variables between subgroups. PLINK version 1.06 was used to obtain summary statistics of the SNPs such as the allele frequency and genotype distribution and to test for Hardy–Weinberg equilibrium (HWE) for each marker based on chi-square test. Association analyses were applied to detect the associations among the SNPs, serological markers, psychological variables, phenotype and surgical endpoints. Multivariate analysis using a logistic regression model was then applied for association analysis. The following variables were included in the multivariate logistic regression model: sex, age at diagnosis, marital status, history of smoking, pANCA, Anti-I2, Anti-OmpC, ASCA IgA, ASCA IgG, CBir1, SNP rs2066845, SNP rs2076756 and SNP rs2066847. Two-sided statistical tests were applied. Odds ratios (OR) and 95% confidence intervals (CI) were estimated. We used a genetic additive model for primary analysis and set a P value = 0.01 as cut point for nominal associations. For the T2 or final outcomes, we assessed all subjects with the outcome of interest (i.e. disease location, disease behaviour, surgery) by that time point regardless of when that outcome was achieved. There was too small a sample size to analyse for predictors of developing those outcomes between T1 and T2 (if not already present by T1).

Ethical Considerations

This study and the Manitoba IBD Cohort Study overall has been approved by the University of Manitoba Research Ethics Board.