EFFICACY and SAFETY of UPADACITINIB in PATIENTS with ACTIVE ANKYLOSING SPONDYLITIS REFRACTORY to BIOLOGIC THERAPY: A DOUBLEBLIND, RANDOMIZED, PLACEBO-CONTROLLED PHASE 3 TRIAL

ABSTRACT

Background: Upadacitinib (UPA) was shown to be safe and effective through 2 years in patients (pts) with active ankylosing spondylitis (AS) naïve to biologic disease-modifying antirheumatic drugs (bDMARDs) in the pivotal phase 2/3 SELECT-AXIS 1 trial.1,2 Objectives: To assess the efficacy and safety of UPA in pts with active AS with an inadequate response (IR) to bDMARDs. Methods: SELECT-AXIS 2 (NCT04169373) was conducted under a master protocol and includes two separate studies (one for AS bDMARD-IR and one for non-radiographic axial spondyloarthritis [nr-axSpA]). The AS bDMARD-IR study is a randomized, double-blind, placebo (PBO)-controlled, phase 3 trial that enrolled adults ≥18 years with AS who met modifed New York criteria, had BAS-DAI and pt's assessment of total back pain scores ≥4 (numeric rating scale 0-10) at study entry, and had an IR to one or two bDMARDs (TNF inhibitor or IL-17 inhibitor). Pts were randomized 11 to receive oral UPA 15 mg once daily (QD) or PBO during the 14-week (wk) double-blind treatment period. The primary endpoint was ASAS40 response at wk 14. Multiplicity-controlled secondary endpoints evaluated at wk 14 were improvements from baseline in disease activity (ASDAS [CRP], ASDAS ID [<1.3], ASDAS LDA [<2.1], BASDAI50, ASAS20, and ASAS PR), pain (total and nocturnal back pain), function (BASFI), objective measure of infammation (SPARCC MRI score of the spine), spinal mobility (BASMI), enthesitis (MASES), and quality of life (ASQoL and ASAS HI). Non-responder imputation incorporating multiple imputation (NRI-MI) was used to handle intercurrent events and missing data for binary endpoints. Cochran-Mantel-Haenszel (CMH) test and mixed-effect model for repeated measures (MMRM) were used for analyzing binary and continuous endpoints, respectively. Treatment-emergent adverse events (TEAEs) assessed through wk 14 are reported for pts who had ≥1 dose of study drug. Results: All 420 randomized pts with active AS received assigned treatment (UPA 15 mg, n=211;PBO, n=209);409 (97%) received study drug through wk 14. Baseline demographic and disease characteristics were generally similar between treatment groups and refective of an active AS bDMARD-IR population (74% male;mean age 42.4 years;mean disease duration 7. 7 years;83% HLA-B27 positive;mean BASDAI 6.8). Signifcantly more pts achieved the primary endpoint of ASAS40 response at wk 14 with UPA vs PBO (45% vs 18%;P<0.0001;Figure 1);UPA showed onset of effect in ASAS40 as early as wk 4 (nominal P≤0.05). All multiplicity-controlled secondary endpoints met statistical signifcance for UPA vs PBO at wk 14 across multiple clinical domains of AS (P<0.0001;Figure 1). The rate of TEAEs was similar between treatment groups through wk 14 (UPA, 41%;PBO, 37%). TEAEs led to discontinuation in 3 (1.4%) pts treated with PBO and none with UPA. Serious infections occurred with UPA (2.4%) but not with PBO and included 4 events of COVID-19 and 1 event of uveitis. Additional events of uveitis were reported in 3 (1.4%) pts treated with PBO. Infammatory bowel disease (IBD) occurred in 1 (0.5%) pt on UPA and none on PBO. No malignancy, major adverse cardiovascular events, venous thromboembolic events, or death were reported with UPA;1 event of malignancy was observed with PBO. Conclusion: UPA 15 mg QD was signifcantly more effective than PBO over 14 wks of treatment in pts with active AS and IR to bDMARDs. No new safety risks were identifed with UPA compared with its known safety profile.3,4 These fndings are consistent with and complementary to those of SELECT-AXIS 1 (bDMARD-naïve AS population),1,2 and support the use of UPA in pts with active AS, including those who had a previous IR to bDMARD therapy.