DEVELOPMENT AND VALIDATION OF AN INTEGRATED RISK SCORE FOR PREDICTING FUTURE RISK OF CROHN'S DISEASE UP TO 7 YEARS BEFORE DIAGNOSIS IN HEALTHY FIRST-DEGREE RELATIVES: THE CCC-GEM PROJECT, A MULTICENTRE PROSPECTIVE COHORT STUDY

Background
Although the cause of Crohn’s disease (CD) is unknown, recent studies have identified a number of biomarkers associated with the risk of developing CD in healthy at-risk individuals. Establishing a combined prediction model that stratifies the future risk of CD in healthy at-risk individuals is the first step towards disease prevention and a better understanding of the preclinical phase of CD.

Methods
We recruited healthy first-degree relatives (FDRs) of patients with CD from 2008-2017 as part of the global multicenter prospective Crohn’s and Colitis Canada Genetic Environmental Microbial (CCC-GEM) Project. After collecting demographic information, blood, urine, and stool samples at recruitment, participants were followed for the development of CD. A GEM-integrative risk score (GEM-IRS), using random survival forest modeling that combined the baseline variables (demographics, measures of gut inflammation – fecal calprotectin (FCP), intestinal barrier function – urinary fractional excretion ratio of lactulose to mannitol (LMR), and fecal microbiome composition and predicted functional capacities – based upon 16S rDNA sequencing and PICRUSt2) to estimate time-to-CD onset, was derived and subsequently validated in two independent testing sets.

Findings
Among 2,619 FDRs followed for a median of 6.8 years, 61 (2.3%) developed CD. The GEM-IRS, developed on the North American Training cohort (n= 1,170), upon validation, showed a c- statistic of 0.789 in the pooled-testing sets (0.786 in the North American Testing cohort (n=1,141) and 0.804 in the Israeli cohort (n=308)). The 4^th quartile group compared to the rest had significantly increased risk of CD onset (hazard ratio [HR] 6.42, 95% confidence interval [CI], 3.10-13.30) in the pooled-testing sets (HR 5.89, 95% CI, 2.70-12.85 in North American Testing cohort; HR 10.7, 95% CI 1.24-92.65 in Israeli cohort). Furthermore, the GEM-IRS predicted CD in pre-specified subgroups of FDRs with minimal subclinical inflammation (fecal calprotectin<50 ug/g) or normal barrier function measures (lactulose/mannitol ratio<0.025), and up to 7 years before diagnosis in the pooled-testing sets. Notably, the top contributors to the risk score are FCP, LMR, superpathway of fucose and rhamnose degradation, fucose degradation pathway, and the genera Holdemania.

Conclusion
The GEM-IRS, which includes biomarkers of gut inflammation, intestinal barrier function, and the gut microbiome, is a valid risk stratification tool for predicting future development of CD in healthy FDRs of persons with CD. The score may be used to guide preventative care for the healthy FDR population.