A combination of clinical, serological, and genetic factors predicts complicated disease course in paediatric-onset Crohn's disease: results from a population-based study
Résumé
Background
Identification of patients at high risk of progressing disease would be invaluable in guiding initial therapy in Crohn’s disease (CD). Clinical parameters at diagnosis are insufficient to predict a disabling course of CD.1 The objective of this study was to evaluate a combination of clinical, serological, and genetic factors to predict complicated disease course in paediatric-onset CD.
Methods
Paediatric-onset CD patients, diagnosed before 17 years between 1988 and 2004 and followed more than 5 years were extracted from the French population-based Epimad registry. Complicated disease course was defined by the evolution from an inflammatory (B1) to a complicated behaviour (stricturing B2, or penetrating B3); or an intestinal resection. Available data included clinical data at diagnosis, serological markers at inclusion (ASCA, ANCA, anti-OmpC, anti-Cbir1, anti-Fla2, anti-Flax) and 370 candidate single-nucleotide polymorphisms (SNP) associated with CD or other immune-mediated diseases or with a role in inflammation pathways, interaction with micro-organisms, or modulation of innate immunity. Lasso logistic regression models with stability selection were used to select variants associated with severe disease. A final logistic adaptive lasso regression model was performed including clinical, serological and selected variants.
Results
A total of 219 patients were included in this study with a median age at diagnosis of 14.3 years (IQR 11.9–16.0). One-third (n = 70) experienced an intestinal resection during the 5 years following diagnosis. Among the 156 patients with inflammatory disease (B1) at diagnosis, about one-third (n = 48) progressed to a complicated behaviour. Final model for resection included complicated behaviour (B2 or B3) at diagnosis, extra-intestinal manifestations at diagnosis and 23 SNPs. Final model for complicated behaviour included ANCA (protective effect) and Asca-IgG (negative effect) levels and 18 SNPs. In both models, half of SNPs were known as susceptibility loci of IBD or CD including NOD2 G908R missense mutation. Six variants were common between the two models. Area under the curve assessed by 5-fold cross-validation were, respectively, 0.84 (IC 95% [0.80–0.88]) and 0.89 (IC 95% [0.87–0.91]).
Conclusions
In this population-based paediatric-onset CD cohort, a combination of clinical, serotypic, and genotypic variables is able to predict disease progression with a high accuracy. After validation in an independent cohort, this prediction score will be helpful to identify patients needing early biological therapy.