173 Comparative effectiveness of secukinumab and golimumab in ankylosing spondylitis assessed by matching-adjusted indirect comparison using pivotal phase III clinical trial data

Abstract

Background: No data are available from head-to-head randomised controlled trials (RCTs) between secukinumab 150mg (an antiinterleukin-17A) and golimumab 50mg (a tumour necrosis factor inhibitor [TNFi]) in patients with active ankylosing spondylitis (AS). Matching-adjusted indirect comparison (MAIC) can be used to estimate comparative effectiveness and enables treatment outcomes to be compared across effectively balanced trial populations. MAIC is an established method in health-technology assessments and the National Institute for Health and Care Excellence has published guidance on appropriate methodology for addressing imbalances in observed covariates between trials. The comparative effectiveness of secukinumab and golimumab up to Week 24 was assessed using MAIC with pooled individual patient data (IPD) from the RCTs MEASURE 1 and MEASURE 2 and published aggregate data from the RCT GO-RAISE. Methods: Pooled MEASURE 1 and 2 data were used to maximise the effective sample size (ESS) for secukinumab. IPD from the secukinumab arms of MEASURE 1 and MEASURE 2 (n=197) were weighted to match the published baseline characteristics of the golimumab arm of GO-RAISE (n=138). Placebo arms were matched in the same way; placebo-adjusted comparisons were possible only until week 16 because patients could receive active treatment from this time onwards. Logistic regression was used to determine weights for age, sex, Bath AS Functional Index, disease duration, C-reactive protein and previous TNFi therapy. Recalculated outcomes from MEASURE 1 and MEASURE 2 (secukinumab, ESS=102; placebo, ESS=81) were compared with data from GO-RAISE (golimumab, n=138; placebo, n=78). Pairwise comparisons reported as odds ratios (ORs [95% CIs]) were performed for Assessment in SpondyloArthritis International Society criteria (ASAS) 20, ASAS 40 and ASAS partial remission (PR) responses at nearest-equivalent timepoints across trials: week 12 (secukinumab)/14 (golimumab), week 14 (golimumab)/16 (secukinumab) and week 24 (secukinumab and golimumab). Non-responder imputation (NRI) was available for all binary outcome data. Strict thresholds were avoided when interpreting P values, in line with American Statistical Association guidance. Results: There was no evidence of differences in ASAS 20 and ASAS 40 responses between secukinumab and golimumab at Weeks 12/14 and 14/16 (both placebo-adjusted). At week 24, non-placebo-adjusted ASAS 20 and ASAS 40 responses using NRI were higher with secukinumab than golimumab (OR [95% CI]: 1.58 [0.93, 2.69], p=0.089 and 1.58 [0.94, 2.64], p=0.084, respectively). There was no evidence of differences in ASAS PR responses between secukinumab and golimumab at weeks 12/14, 14/16, and 24. A sensitivity analysis conducted after adding Bath AS Disease Activity Index score to the matching parameters yielded similar results. Conclusion: There was no evidence of differences in ASAS responses between secukinumab and golimumab in placebo-adjusted analyses. In non-placebo-adjusted analyses, secukinumab showed higher ASAS 20 and ASAS 40 responses than golimumab at week 24.