A12: The Role of Serum S100A12 Protein Levels in Disease Flare After Withdrawal of Anti‐tumor Necrosis Factor Therapy in Polyarticular Forms of Juvenile Idiopathic Arthritis

Abstract

Background/Purpose:Anti‐TNF therapy for polyarticular forms (extended oligo‐, rheumatoid factor +/− polyarthritis) of JIA (PF‐JIA) results in up to 50% of patients (pts) demonstrating clinically inactive disease (CID). This study determined the pattern of serum S100A12 levels at the time of withdrawal of anti‐TNF therapy.Methods:In 16 centers, 137 pts with PF‐JIA in CID on anti‐TNF therapy were enrolled and followed for at least 14 months (mos). During the first 6 study mos pts were maintained on anti‐TNF therapy and if CID was maintained, then anti‐TNF therapy was stopped. Background medications were stable. S100A12 levels were obtained at the time of anti‐TNF withdrawal. The primary outcome was disease “flare” defined by worsening of ≥30% in ≥ 3 of the 6 core set variables, with no more than 1 improving by ≥30%. Parameters had to increase by at least the following amounts: MD and parent globals by 2 units, active and limited joints by 2, CHAQ by 0.125 and ESR from normal to abnormal.Results:24 pts failed to maintain CID in the first 6 study mos and 7 pts were discontinued from the study for other reasons. 106 pts were available for this analysis of whom 39 (37%) experienced flare. The S100A12 levels at time of anti‐TNF withdrawal did not differ significantly in regards to JIA type, presence of ANA, MTX co‐therapy, or type of anti‐TNF therapy and did not correlate with previous duration of CID. Globally, S100A12 at the time of withdrawal did not differ significantly according to disease flare (flare 73 +/− 117 ng/ml, no flare 80 +/− 220 ng/ml) (median ++/− standard deviation). Receiver‐operating curve (ROC) analysis computing S100A12 at time of anti‐TNF withdrawal against flare for the entire study period demonstrated an area‐under‐the‐curve (AUC) of 0.51, 95% confidence interval (CI) 0.39–0.62, for prediction of flare within 60 days AUC 0.66, 95% CI 0.56–0.75, for prediction of flare within 90 days AUC 0.64, 95% CI 0.54–0.74 and for prediction of flare within 120 days AUC 0.54, 95% CI 0.44–0.64. The S100A12 level at time of withdrawal correlated inversely with the time to flare (Spearman rank correlation = −0.34, p = 0.05). Flares occurred earlier in pts with predefined high (>120 ng/ml) vs. low (≤120 ng/ml). S100A12 levels at time of withdrawal among pts who flared (median time to flare 36 vs. 114 days, p = 0.02). The overall flare rate in patients with high vs. low S100A12 levels at time of withdrawal was 35% vs. 38%, respectively. Kaplan‐Meier analysis of disease flare in pts with high vs. low S100A12 levels at time of withdrawal also demonstrated a trend towards earlier disease flare in pts with high S100A12 levels (log rank test significance 0.07).Conclusion:In this prospective study, a substantial proportion of pts with PF‐JIA experienced disease flare after anti‐TNF withdrawal. Serum S100A12 levels at time of anti‐TNF withdrawal did not differ between pts subsequently experiencing disease flare and those not experiencing flare throughout the entire study period. However, pts with high S100A12 levels (>120 ng/ml) experienced earlier disease flare.