P319 Accurate cut-offs for predicting endoscopic activity and mucosal healing in Crohn's disease with fecal calprotectin

Abstract

Background: The assessment of symptoms has not proved to be useful in accurately establishing endoscopic activity in Crohn's disease (CD). Fecal calproctectin (FC) has demonstrated high precision to detect endoscopic activity in CD. However, the clinical applicability of FC is limited as no consensus exists on the optimum cut-off point for establishing endoscopic activity and mucosal healing. The aim of this study is to analyze whether FC is a good tools for generating highly accurate scores for the prediction of the state of endoscopic activity and mucosal healing. Methods: The simple endoscopic score for Crohn's disease (SES-CD) and the Crohn's disease activity index (CDAI) was calculated for 71 patients diagnosed with CD. FC was measured by the Enzyme-Linked Immunosorbent Assay test. An accuracy analysis was made by estimating ROC curve of FC with respect to the SES-CD. We calculate both CF cut-off point with greater accuracy to establish endoscopic activity and as values for sensitivity and specificity, as well as for predictive scores: positive and negative, global accuracy and likelihood ratios (LR). It was calculated that a cut-off with specificity >90% and LR+ ≥10 would present strong evidence to support the diagnostic hypothesis (endoscopic activity) and a cut-off point with sensitivity >90% and LR- ≤0.1 would offer strong enough for to reject the hypothesis. Finally, the Fagan nomogram was calculated to determine the probability of endoscopic activity or mucosal healing after obtaining the biomarker score and according to clinical symptoms. Results: A FC cut-off of 170 μg/g (Sensitivity 77.6%, Specificity 95.5% and LR+ 17.06) predicts a high probability of endoscopic activity, and a FC cut-off of 71 μg/g (Sensitivity 95.9%, Specificity 52.3% and LR- 0.08) predicts a high probability of mucosal healing. In our sample, the prevalence of endoscopic activity was 69%, but if a patient shows FC ≥170 μg/g they have a 97% probability of presenting endoscopic activity; and if they have FC ≤71 μg/g the probability would be 84% for presenting mucosal healing. Clinical symptoms modified the probabilities of predicting endoscopic activity (100% if clinical activity vs 89% if clinical remision) or mucosal healing (75% if clinical activity vs 87% if clinical remision) in the diagnostic scores generated. Conclusions: FC is a useful tool for generating highly accurate scores for predicting the state of endoscopic activity or mucosal healing in CD patients. Although, it is important to take into account the specific clinical context in order to interpret the probabilities of presenting endoscopic activity or mucosal healing according to the FC level.