P789 Inflammatory Bowel Disease subgroups identified by longitudinal faecal calprotectin profiles: A 10-year retrospective analysis of the Lothian IBD Registry

Constantine-Cooke, N.(1,2)*;Monterrubio-Gómez, K.(1);Plevris, N.(3);Derikx, L.A.A.P.(4);Lyons, M.(3);Jenkinson, P.W.(3);Gros, B.(3);Jones, G.R.(5);Vallejos, C.A.(1,6);Lees, C.W.(2,3);

(1)University of Edinburgh, MRC Human Genetics Unit, Edinburgh, United Kingdom;(2)University of Edinburgh, Centre for Genomic and Experimental Medicine, Edinburgh, United Kingdom;(3)Edinburgh IBD Unit, Edinburgh IBD Unit, Edinburgh, United Kingdom;(4)Radboud University Medical Centre, Gastroenterology and Hepatology, Nijmegen, The Netherlands;(5)University of Edinburgh, Centre for Inflammation Research, Edinburgh, United Kingdom;(6)The Alan Turing Institute, The Alan Turing Institute, London, United Kingdom;

Background

Faecal calprotectin (FCAL) is routinely used to monitor mucosal inflammation, and high FCAL values are associated with poor outcomes in IBD.

Having previously found clusters characterised by distinct FCAL profiles in a Crohn’s disease (CD) cohort of 356 subjects, we aimed to extend this approach to all IBD patients treated by the Edinburgh IBD Unit.

Methods

Subjects were required to have a diagnosis of IBD, three FCAL measurements within ten years of diagnosis between January 2005 and September 2019, and be treated by the Edinburgh IBD Unit.  All FCAL was measured on the same ELISA essay as part of routine clinical care.

Latent class mixed models, an extension of linear mixed effects models with cluster specific fixed effects, were used to model and cluster FCAL trajectories. Fixed and random effects were modelled using natural cubic splines, which allows flexible models with relatively low computational cost when compared to high degree polynomials. Akaike information criterion, Bayesian information criterion and alluvial plots were used to decide on the appropriate number of clusters. 

CD subjects were modelled independently of UC and IBDU subjects, with similar clusters across disease subtypes given the same numbering for ease of comparison. Hereafter, we will describe the UC and IBDU patients collectively.

Results

1,214 CD and 1,194 UC/IBDU subjects met the inclusion criteria. Across follow-up, 10,503 (median 7 per subject) FCAL measurements were available for patients with CD and 7,495 (median 5 per subject) for those with UC/IBDU.


Figure 1. Log-transformed ten-year faecal calprotectin profiles for Crohn’s disease subjects. Red curves demonstrate overall profile for each cluster. The blue dotted lines indicate an FCAL of log (250 μg/g), the commonly accepted threshold for biochemical remission.

For CD, six clusters were found to be appropriate, whereas five clusters were found to be the most appropriate for UC/ IBDU.

The overall cluster prmofiles were remarkably similar between CD and UC/IBDU.  Both CD cluster 2 and UC cluster 2, the largest clusters, were characterised by consistently high FCAL across follow-up: remaining above the 250 μg/g threshold commonly used for biochemical remission (blue dotted lines in the figures).



Figure 2. Log-transformed ten-year faecal calprotectin profiles for ulcerative colitis subjects

Conclusion

After grouping CD and UC/IBDU subjects in a large cohort by their 10-year FCAL trajectories, we found remarkably similar trajectories across the two groups. In both cases, we found a large cluster of subjects with consistently elevated FCAL, which warrants further investigation.