Objectives: We aimed to identify different anti-TNF-α agents for ankylosing spondylitis (AS) assessed in randomised controlled trials (RCTs) and to compare them within a single evidence synthesis framework.
Methods: A Bayesian network analysis method was used to generate direct and indirect comparisons while maintaining randomisation. The main outcomes were the proportion of ASAS20 patients at the follow-up of 12 weeks. The analysis was made based on an intention-to-treat basis.
Results: Data were combined from 14 (RCTs) that included 17833 patients randomised to 7 treatment strategies, including placebo. Except for 3mg/kg infliximab at 0, 2, 6 weeks, all other treatments were demonstrated to be more effective than placebo in the terms of clinical index ASAS20. Compared with 25 mg etanercept twice a week, 50 mg etanercept once a week, 50 mg golimumab, 100 mg golimumab every four weeks, 5mg/kg infliximab at 0, 2, 6 weeks and 40 mg adalimumab every other week for 12 weeks seemed to be more effective (odds ratios [OR] 1.38, 1.22, 1.26, 1.29, 1.38, and 1.25, respectively), while etanercept 50 mg twice a week have the similar efficacy (odds ratios [OR] 1.08), and infliximab 3 mg/kg at 0, 2, 6 weeks was less effective (odds ratios [OR] 0.69). However, all of these between-treatment comparisons detected no significant analysis. Finally, ranking analysis suggested that infliximab 5 mg/kg at 0, 2, 6 weeks may be the best efficacious therapy.
Conclusions: Our results suggested that infliximab 5 mg/kg at 0, 2, 6 weeks seems to be the best efficacious therapy, while infliximab 3 mg/kg at 0, 2, 6 weeks maybe could not be considered in the future studies. Future studies could pay more attention to the comparison of different anti-TNF agents, instead of comparison between anti-TNF agents and placebo.