Background Although reducing inflammation has been associated with pain improvement, the two do not always correlate. Recent studies have suggested that, in addition to its role in inflammation pathogenesis, IL-23 may be involved in pain regulation in a lymphocyte-independent manner1. Guselkumab (GUS), a fully human monoclonal antibody that selectively inhibits IL-23, has demonstrated safety and efficacy in treating multiple domains of active PsA in the DISCOVER-1&2 (D1&D2) trials2,3.

Objectives To quantify the role of reducing inflammation on the observed relationship between GUS and pain response in PsA patients (pts) using mediation modelling.

Methods Pooled data from the D1&D2 studies were analyzed. Pts in D1 had ≥3 swollen and ≥3 tender joints (SJC/TJC) and C-reactive protein (CRP)≥0.3 mg/dL; in D2, pts had ≥5 SJC and ≥5 TJC and CRP≥0.6 mg/dL. 31% of D1 pts received 1-2 prior tumor necrosis factor inhibitors (TNFi); D2 pts were bio-naïve. Pts were randomized 1:1:1 to GUS 100 mg every 4 weeks (Q4W); GUS 100 mg at W0, W4, then every 8 weeks (Q8W); or placebo (PBO); PBO pts crossed over to GUS 100 mg Q4W at W24. Pts with history of fibromyalgia were excluded from the analysis. Least square mean changes in pt-reported pain (0-100 VAS) through W52 were estimated with a repeated measures linear mixed model adjusting for known pain determinants. Mediation modelling was performed separately for Q4W & Q8W, W4 & W24, and TNFi-naïve & -experienced (exp) pts. In each model, change in pt-reported pain was the dependent variable; treatment regimen was the independent variable; inflammation, measured by change in SJC or CRP, was the designated mediator; covariates were: age; sex; and baseline (BL) pain score, BMI, SF-36 MCS score, and NSAID use.

Results Mean (SD) BL pain levels in the GUS Q4W, GUS Q8W, and PBO groups were 60.4 (19.8), 62.0 (20.2), and 61.1 (19.6), respectively. Treatment with GUS was associated with significantly greater pain improvement compared with PBO as early as W4 (ΔQ4W-PBO [95%CI]: -4.9 [-7.6, -2.2]; ΔQ8W-PBO [95%CI]: -5.2 [-7.9, -2.5] (Figure 1). These between-group differences were further enhanced by W24 (ΔQ4W-PBO [95%CI]: -14.6 [-17.6, -11.5]; ΔQ8W-PBO [95%CI]: -14.3 [-17.3, -11.2]); by W52, GUS-randomized pts exhibited an approximate 30-point (̴50%) decrease in pain. Similar results were observed for TNFi-naïve and TNFi-exp pts.