P470 Early change in faecal calprotectin during anti-TNFα initiation is strongly predictive of outcome in Crohn’s disease
P. Pavlidis*, G. Chung- Faye, P. Dubois, I. Bjarnason, B. H. Hayee
King’s College Hospital, Department of Gastroenterology, London, United Kingdom
Anti-TNFα agents represent a significant advance in the management of Crohn’s disease (CD). There remains, however, a significant proportion of patients that will not respond and the challenge remains to detect or predict these as early as possible. A pilot study (Leal et al, Gut 2015) to detect prognostic markers of early response to anti-TNFα therapy identified the 2 genes coding for the calprotectin subunits (S100A8 and S100A9) to be amongst the most highly expressed gene transcripts in non-responders. We hypothesised that measuring calprotectin in faeces (FCAL) pre- and post-induction in patients with active inflammatory luminal CD could be used to predict response to anti-TNFα therapy.
Patients were identified through our prospectively kept electronic IBD database. Only anti-TNFα naïve patients with non-penetrating disease (B1 or B2) who received standard induction with infliximab (IFX; 0, 2, and 6 wk, 5mg/kg) or adalimumab ADA (160/80mg) respectively, between January 2013 to December 2014, were included. FCAL measurements (Buhlmann ELISA, values mcg/g faeces) were performed at planned time points pre- and post-induction (within one month before treatment then 68– weeks post first ADA dose, 81–0 weeks post first IFX dose). Outcomes were assessed at 6 and 12 months based on clinical activity scores and the use of corticosteroids (remission [RM] Harvey–Bradshaw Index [HBI] < 5, off corticosteroids > 2 months; response [RS] drop in HBI > 3, off corticosteroids; or non-response [NR]). Outcome measures were defined before data analysis.
At 6 months, RS = 23 (72%) (median HBI 4 [1, 10]; FCAL 55 [10, 1696]); and RM = 17 (53%) (HBI 3 [1, 4] and fcal 42 [10, 171]). In NR (28%) median HBI was 8(7, 9), FCAL 355 (103, 1 400). There was a significant difference in the ΔFCAL from baseline to post-induction in the 3 groups (p < 0.0001). Comparing NR to RM+RS, the area under the curve (AUC) of ΔFCAL to predict outcome at 6 months was 0.97. Using receiver-operating characteristic curve (ROC) analysis, a Δ70% returned a sensitivity and specificity for response of 99% and 96%, respectively (likelihood ratio [LR] = 23). At 12 months, 13/32 (41%) were in remission, but ΔFCAL did not perform as well for this time point (AUC 0.65, p = 0.17). However, a post-induction FCAL<170 had a sensitivity and specificity for remission of 68% and 92%, respectively (LR = 8.9).
When assessed immediately after anti-TNF induction, the change in FCAL (Δ70%) from baseline accurately predicts 6-month outcomes in CD, whereas a FCAL<170 mg/g post-induction predicts remission at 12 months. This approach could be used to guide treatment choices and early decision making.