AG

AA

bst

ract

s

a Including singletons b Rare variant observed in only 1 out of 102 alleles

Imputation performed using BEAGLE with default parameters and an AJ-specific referencepanel of 51 samples. r2 represents estimated correlation between imputed and true genotypes.

Tu1904

Towards Individualised Risk Prediction for Crohn's DiseaseKirstin Taylor, Natalie J. Prescott, Simon H. Anderson, Peter M. Irving, Sarah L. West,Daniel J. Crouch, Christopher G. Mathew, Jeremy D. Sanderson, Cathryn Lewis

Introduction: Although the relative risk (RR) of developing Crohn's disease (CD) is increasedin first degree relatives (FDR) of probands by 13-27 fold,(1) it is difficult to apply this onan individual basis. Recently, risk models for complex disease using information fromgenome-wide association and epidemiological studies have been built. Potentially, suchmodels could identify FDR at highest risk of CD, for screening and early intervention. Aim:To generate a risk model for CD, and ultimately determine if such a model can predictpresymptomatic or future development of CD. Method: CD patients were contacted toidentify asymptomatic FDR aged 18-55 who wished to participate in research into riskprediction and presymptomatic diagnosis of CD. (Recruitment target of 500 FDR.) Demo-graphics, medical and smoking history were recorded in FDR, and a saliva sample wassupplied for DNA extraction. Genetic analysis used the Immunochip (Illumina), and geno-types were determined for disease susceptibility loci established by CD genome-wide meta-analysis.(2) A multiplicative risk model was generated using risk prediction software(REGENT)(3) based on genotype and smoking status, and CD RR determined for each FDR.Results: The first 100 FDR have been recruited. Demographics: 53% siblings, 36% offspring,11% parents. Mean age 37, females 66%, smokers 25%, ex-smokers 33%. NOD2 genotypecounts were not significantly different to their frequencies in the control population, butwere significantly different to cases (13% heterozygous, 0% homozygous, p=0.64 for controlsvs FDR, p=0.001 for cases vs FDR). CD RR ranged from 0.04 - 10.07 (median 0.96) andwas significantly higher than that in the general population (figure 1): 19% of FDR had aRR >2, compared to an expected proportion of 11% (p=0.02). Comparison of highest andlowest RR quartiles (FDR) revealed similar demographics between the 2 groups. RR for thesequartiles is shown in figure 2. More FDRs in the highest RR quartile had > 2 CD-affectedrelatives compared with FDRs in the lowest RR quartile but this was not significant (5 vs1, p=0.19). Conclusion: Increased RR of CD is confirmed in FDR compared to the generalpopulation. Risk stratification is possible, but follow-up is required to determine if suchrisk scores can be used to predict presymptomatic CD or future development of CD.Additionally, risk awareness may encourage smokers to quit. Once recruitment to the studyis completed, FDR in the highest and lowest risk quartiles will proceed to further testsincluding faecal calprotectin, serum antibodies and capsule endoscopy, and will undergofollow-up for 10 years. References: 1.Lewis CM et al. J Med Genet 2007 44:689-6942.Franke A et al. Nat Genet 2010 42:1118-1125 3.http://cran.r-project.org/web/packages/REGENT/index.html

S-874AGA Abstracts

Tu1905

Interaction of IBD-Associated Genetic Polymorphisms and the Microbiome ofthe Pelvic PouchAndrea D. Tyler, Natalie Berard, David S. Guttman, Denis O. Krause, Mark S. Silverberg

Introduction: Ileal inflammation affecting the pelvic pouch following ileal pouch-anal ana-stomosis (IPAA) in patients with ulcerative colitis (UC) is thought to develop in individualswith a genetic predisposition and changes in the gut-associated microbiome. This modelcan be used to study the relationship between these genetic polymorphisms and the ilealmicrobiome. The aim of this study was to evaluate the role of well-studied inflammatorybowel disease (IBD) genetic polymorphisms in influencing the composition of the microbiomein the pouch following IPAA. Methods: Patients with confirmed UC or familial adenomatouspolyposis with IPAA, were recruited from Mount Sinai Hospital in Toronto. Biopsies takenfrom the afferent limb and pouch were snap frozen. Peripheral blood was collected in anEDTA vacutainer. Microbial DNA was extracted and sequenced using primers for the V1-V3 hypervariable regions of microbial 16S rRNA genes on the 454 GS FLX TitaniumSequencing system. These sequences were taxonomically assigned to bacterial genera bycomparing them to the SILVA reference database using the mothur computational microbialecology software package, with a minimum bootstrap cutoff of 60%. Resulting genera datawas dichotomized based on whether or not corresponding sequences were detected. Thosegenera which were detected in fewer than 5 samples were excluded. Genotyping wasperformed using the Sequenom iPLEX platform with 66 SNPs selected based on previousIBD and pouchitis genetics studies. Correlation between SNPs and microbiome were assessedseparately in the pouch and afferent limb using logistic regression with an additive geneticmodel. False discovery rate correction for multiple testing was applied to correct for multiplecomparisons. Results: In total, samples from 72 patients were included in the analysis. 54SNPs and 81 genera met quality control measures and were included. Variants at 8 geneticloci were associated with microbial taxa in the pouch (puncor<0.001) (Table 1). The SNPrs2076756 in NOD2 was associated with Citrobacter (pouch and afferent limb) and anuncultured member of the family Erysipelotrichaceae (pouch only) (puncor=2.7x10-4 andpuncor=0.003, respectively). No further associations were observed in the afferent limb. Noneof these associations withstood correction for multiple testing. Conclusion: IBD-associatedgenetic polymorphisms are potentially associated with specific microbial taxa in the pelvicpouch. Corroboration of these results is required in a larger sample size. Identification of