External validation of serum biomarkers predicting short-term and mid/long-term relapse in patients with Crohn's disease stopping infliximab.

Abstract

Objective: In patients with Crohn's disease (CD) on combination therapy (infliximab and immunosuppressant) and stopping infliximab (cohort from the study of infliximab diSconTinuation in CrOhn's disease patients in stable Remission on combined therapy with Immunosuppressors (STORI)), the risk of short-term (≤6 months) and mid/long-term relapse (>6 months) was associated with distinct blood protein profiles. Our aim was to test the external validity of this finding in the SPARE cohort (A proSpective Randomized Controlled Trial comParing infliximAb-antimetabolites Combination Therapy to Anti-metabolites monotheRapy and Infliximab monothErapy in Crohn's Disease Patients in Sustained Steroid-free Remission on Combination Therapy).

Design: In SPARE, patients with CD in sustained steroid-free clinical remission and on combination therapy were randomly allocated to three arms: continuing combination therapy, stopping infliximab or stopping immunosuppressant. In the baseline serum of the STORI and SPARE (arm stopping infliximab) cohorts, we studied 202 immune-related proteins. The proteins associated with time to relapse (univariable Cox model) were compared between STORI and SPARE. The discriminative ability of biomarkers (individually and combined in pairs) was evaluated by the c-statistic (concordance analysis) which was compared with C-reactive protein (CRP), faecal calprotectin and a previously validated model (CEASE).

Results: In STORI and SPARE, distinct blood protein profiles were associated with the risk of short-term (eg, high level: CRP, haptoglobin, interleukin-6, C-type lectin domain family 4 member C) and mid/long-term relapse (eg, low level: Fms-related tyrosine kinase 3 ligand, kallistatin, fibroblast growth factor 2). At external validation, the top 10 biomarker pairs showed a higher c-statistic than the CEASE model, CRP and faecal calprotectin in predicting short-term (0.76-0.80 vs 0.74 vs 0.71 vs 0.69, respectively) and mid/long-term relapse (0.66-0.68 vs 0.61 vs 0.52 vs 0.59, respectively).

Conclusion: In patients with CD stopping infliximab, we confirm that the risk of short-term and mid/long-term relapse is associated with distinct blood protein profiles showing the potential to guide infliximab withdrawal.

Trial registration number: NCT00571337 and NCT02177071.