P-194 YI A Model Integrating Genetic and Environmental Factors Can Differentiate Crohnʼs Disease from Ulcerative Colitis

Abstract

Crohn's disease (CD) and ulcerative colitis (UC) are 2 related yet different forms of chronic inflammatory bowel disease (IBD). Clinically, it is relevant to differentiate these 2 forms of disease and adopt specific treatment modalities. Discriminative tools combining different bio-molecular markers have been proposed but there is a lack of replication in independent cohorts. The aim of this study was to derive a model integrating genetic markers and environmental factors to differentiate CD from UC using a discovery cohort and an independent replication cohort. The Wellcome Trust Case-Control Consortium (WTCCC) IBD GWAS (1,748 CD and 2,361 UC; non-Jewish, European ancestry) served as a discovery cohort to identify genetic variants which were differentially associated with CD versus UC. The CC/UP cohort, an ongoing IBD GWAS (692 CD and 566 UC; non-Jewish, European ancestry) served as an independent replication cohort. In the CC/UP cohort, questionnaires were used to assess demographic and environmental factors including ethnicity, history of appendectomy (prior to IBD diagnosis), and tobacco use (classified as current smoker versus past smoker or never smoker at the time of IBD diagnosis). History of appendectomy and current smoking status prior to the IBD diagnosis were compared between CD and UC. Discriminative accuracy was evaluated using the area under the receiver operating characteristic curves (AUC). AUCs of different models were statistically compared through the SAS 9.2 GPLOT procedures. In the WTCCC discovery cohort, comparing CD versus UC, there were 121 single nucleotide polymorphisms (SNP) with nominal P values less than 10E-05. To avoid collinearity between markers, stepwise backward procedure (threshold of P < 0.01) was adopted and selected 79 SNPs into the model [AUC: 0.91, 95% CI, ±0.004]. We applied the same model of 79 SNPs in the independent CC/UP cohort and found an AUC of 0.74 [95% CI, ±0.01]. Furthermore, in the CC/UP cohort, current smoking and appendectomy prior to IBD diagnosis were significantly associated with CD versus UC (Odds ratios (OR) = 6.12 [3.79–9.88], P < 10E-07, and OR = 3.03 [1.48–6.23], P = 0.002, respectively). A model integrating genetic markers and environmental factors (current smoking and appendectomy before the diagnosis of IBD) had a discriminative accuracy between CD versus UC with an increase in AUC from 0.74 to 0.78 [95% CI, ±0.01] (P < 10E-06). Through 2 independent cohorts, we identified a model integrating genetic markers (79 SNPs) and environmental factors (current smoking and history of appendectomy before IBD diagnosis) with modest discriminatory power to differentiate CD from UC [AUC: 0.78]. Potential clinical applications of this work include applying genetic and environmental risk modeling to predict complicated disease outcomes and to potentially guide disease-specific treatment choices.