Comparison of the Efficacy and Safety of Myasthenia Gravis Treatments: A Bayesian Network Meta-Analysis

Structured Abstract Importance: Many novel targeted therapies have shown promise in treating generalized myasthenia gravis (MG); however, no head-to-head prospective studies have compared efficacy and safety between one another, nor to existing therapies. Objective: We conducted a nuanced comparative analysis of MG treatments that accounts for differences in trial designs and populations. Data Sources and Study Selection: We searched Scopus, Embase, Web of Science, CENTRAL, and MEDLINE up to May 2, 2025, for randomized trials for MG. We included prospective multi-arm studies of adults (18 or older) with generalized MG that reported mean treatment difference, with uncertainty, in MG-Activities of Daily Living (MG-ADL) or Quantitative MG (QMG) score change from baseline. Of 4,823 eligible studies, 32 studies (27 placebo-controlled) met these criteria. Data Extraction and Synthesis: Two independent reviewers extracted data in accordance with PRISMA guidelines and assessed risk of bias. Data was pooled with a Bayesian random-effects model. Main Outcomes and Measures: The primary outcome was the difference in MG-ADL or QMG (patient-reported and physician-reported outcome measures, respectively) change from baseline between treatment arms. Results: Twenty-seven trials (placebo n=1,086; treatment n=1,232) were included that targeted B cells, neonatal Fc receptors (FcRn), complement activity, CD40, and interleukin-6 signaling as well as immunoglobulin G and broad immunosuppression therapy. QMG scores decreased by an average of -3.6 (95% Credible Interval: [-4.40, -2.72]) points more in patients treated with FcRn inhibitors than those treated with placebo and standard of care, followed by -2.59 [-3.93, -1.27] points for C5 complement inhibitors (C5i), and -2.50 [-5.15, 0.13] points for CD19+ B-cell depletion therapy (BCDT). Sensitivity analyses detected potential confounding of treatment effects by age, sex, and baseline MG-ADL and QMG scores. Therefore, differences in trial populations may mask the treatment effect of CD20+ BCDT. Patients treated with FcRn inhibitors had greater odds of treatment-related adverse events (Odds Ratio: 2.20 [1.52, 3.38]) while C5i and CD19+ BCDT showed comparable odds to placebo and standard of care. Conclusion and Relevance: FcRn inhibitors, C5i, and CD19+ BCDT have comparable efficacies in treating MG. However, our findings highlight how differences between trials may confound the interpretation of study outcomes, warranting further prospective comparative studies. Key Points Question: How do the different therapies for myasthenia gravis (MG) compare in terms of their efficacy and safety? Findings: In this Bayesian network meta-analysis, we find comparable efficacies between C5 complement inhibitors, FcRn inhibitors, and CD19+ B-cell depletion therapy. We also demonstrate how confounders can mask or amplify treatment effects. Meaning: This study serves as a cautionary tale against uncontextualized comparisons of treatment efficacy and warrants prospective studies that are better suited for determining optimal MG treatment strategies.