Anti-TNF withdrawal in IBD: Relapse and recapture rates and predictive factors from 160 patients in a pan-UK study

Introduction The mechanisms behind primary non-response (PNR) to infliximab (IFX) in IBD are still incompletely understood. The role of IFX trough levels (TL) and early antibody formation (ATI) during the induction phase (0-2-6 weeks) are contradicting. Furthermore, the evolution of serum tumor necrosis factor (TNF) during IFX induction has been sparsely studied. Aim We investigated if serum markers of inflammation or drug exposure help in understanding what is driving PNR. Methods In this retrospective single-center study, we identified a cohort of 201 anti-TNF naive Crohn's disease (CD) patients who received IFX induction and had serum samples drawn at weeks 0, 2, 6 and 14. In all samples CRP, albumin, TNF (Prometheus Laboratories Inc.), ATI (measured with drug tolerant homogeneous mobility shift assay, Prometheus Laboratories Inc.) and TL (in-house-developed ELISA) were assessed. PNR was defined as complete absence of clinical improvement at week 14 (physician global assessment). All measurements were compared between responders and those with PNR. Results The incidence of PNR was 8% (n=16). In univariate analysis, low albumin at w6 was associated with PNR (P=0.01, Mann Whitney test). We observed a significant increase of serum TNF after each IFX infusion with medians at w0 and w14 of 1.6 pg/ml (IQR 0.9-2.7) and 7.7 pg/ml (IQR 4.5-11.5) respectively (P<0.0001, KruskalWallis test) (see figure 1). In patients with PNR, this rise in TNF (w14-w0) was significantly lower than in responders (P=0.03, Mann Whitney test). In an attempt to classify inflammation driven by TNF relative to the overall inflammation, we compared TNF/CRP ratio between both groups. A stepwise multiple logistic regression model identified albumin at w6 and TNF/CRP at w0 to be independent significant predictors (P<0.01) of PNR to infliximab at w14, with OR (95% CI) of 0.08 (0.02-0.37) and 3.10 (1.54-6.25) respectively. Conclusions A high disease burden (represented by low albumin, high CRP and serum TNF) and not IFX TL or ATI (detectable in 21% of patients at w14) are the most important factors driving PNR to IFX. TNF and CRP separately did not predict primary response but a higher TNF/ CRP ratio before start of IFX was predictive for PNR, contradicting the theorem that PNR might be due to 'non-TNF-driven' disease. Although the median TL at w14 between both groups did not differ significantly (P=0.48), this ratio indicates that the contribution of TNF in inflammation might even be higher in PNR before IFX start, and possibly demanding a higher loading dose. We further observed an increase of serum TNF after each IFX infusion and this was less in PNR. The mechanism behind this increase remains unclear. These results warrant further investigation for the role of disease burden in PNR to IFX.