To synthesize all randomized evidence on complement-targeted therapies in severe Guillain-Barré syndrome using Bayesian meta-analysis, and to determine whether anti-C5 and anti-C1q inhibition differ in efficacy when structural confounding by IVIg co-administration is accounted for.

Standard treatment for severe Guillain-Barré syndrome has not changed in 30 years, and approximately 20% of patients remain unable to walk at 6 months. Eculizumab (anti-C5) and ANX005/tanruprubart (anti-C1q) have been tested in randomized placebo-controlled trials with discordant results, and no prior synthesis has integrated all available trials.

We searched PubMed, Embase, and Scopus through May 2026 for randomized double-blind placebo-controlled trials of complement inhibitors in adults with severe GBS. Three pre-specified pools were analysed using Bayesian random-effects meta-analysis (bayesmeta) with weakly informative priors: eculizumab + IVIg vs placebo + IVIg at week 4 (Pool A1) and week 24 (Pool A2), and ANX005 monotherapy vs placebo at week 8 (Pool B1). Frequentist Hartung-Knapp-Sidik-Jonkman analyses were fitted as cross-checks. Risk of bias was assessed with Cochrane RoB 2 and certainty of evidence with GRADE.

Four trials enrolling 382 patients were included. In Pool A1, the pooled risk ratio was 0.98 (95% credible interval, 0.52–1.94; posterior probability of benefit, 47%). Pool A2 was not interpretable owing to substantial heterogeneity (I², 81%; Q p = 0.021). In Pool B1, the pooled odds ratio was 2.17 (95% credible interval, 0.94–4.76; posterior probability of benefit, 96%; I², 0%). Drug class was perfectly confounded with IVIg co-administration. GRADE certainty was low for the two interpretable pools and very low for Pool A2.

Anti-C1q monotherapy was associated with probable benefit on short-term outcomes. Anti-C5 added to IVIg was not. A randomized trial of tanruprubart against IVIg in a Western AIDP-predominant cohort is needed.