Safety of C5-targeted therapies in generalized myasthenia gravis: a systematic review and meta-analysis of randomized controlled trials with long-term extension data

Title Safety of C5-targeted therapies in generalized myasthenia gravis: a systematic review and meta-analysis of randomized controlled trials with long-term extension data Summary This systematic review and meta-analysis estimates the safety of complement C5-targeted therapies in adults with generalized myasthenia gravis (gMG). Eligible studies are randomized, placebo-controlled trials of C5-targeted monotherapy added to stable background therapy, for example eculizumab, ravulizumab, zilucoplan, cemdisiran, pozelimab, gefurulimab or any other agent acting on C5. The open-label extensions of these trials are used to estimate long-term, exposure-adjusted safety. The primary outcomes are serious adverse events, discontinuation due to adverse events, serious infections and all-cause death during the randomized double-blind period. Secondary outcomes include: any, severe and treatment-related adverse events specific infections, including meningococcal infection common adverse events and injection-site reactions disease-related harms: clinical deterioration, myasthenic crisis, rescue therapy and hospitalization efficacy: MG-ADL, QMG and MGC scores The long-term outcomes are meningococcal infection and death per 100 patient-years, and the proportion of patients with at least one serious adverse event. Binary outcomes are pooled as Mantel–Haenszel risk ratios and risk differences, with random-effects models (REML with Hartung–Knapp adjustment) as sensitivity analyses. Continuous outcomes are pooled as mean differences. Risk of bias is assessed with RoB 2 and certainty of evidence with GRADE. 1. Administrative information Registry: Open Science Framework (OSF). Review type: systematic review and pairwise meta-analysis of randomized controlled trials, with a descriptive and pooled analysis of open-label extensions. Reporting guidelines: PRISMA 2020 for the review; PRISMA-P 2015 for the protocol; the PRISMA harms extension where applicable. Authors: Aasim Ali, Aswad Mustafa, Muskan Ismail, Zia Ur Rehman, Sameen Sarfraz, Ahmad Butt, Salman Riaz, Mahad Bin Yasir, Muhammad Abdullah Ali, Muhammad Abdullah, Armaghana Abdullah, Mukesh Kumar Sharma. Affiliations: Neurology Department, Allied Hospital Faisalabad, Pakistan; Neurology Department, Dhankuta District Hospital, Nepal. Guarantor: Aasim Ali. Contributions: AA conceived the review, designed the methods and conducts the analyses. AM and MI conduct the searches and screening. ZUR, SS and AB extract data and assess risk of bias. SR, MBY, MAA, MA and ArA verify data and references. MKS supervises the work and provides methodological guidance. Amendments: any change after registration is recorded on OSF with its date, description and rationale, and is reported in the final article. Funding: none. Competing interests: none declared. 2. Introduction 2.1 Rationale C5 inhibition is an established treatment class for acetylcholine-receptor-antibody-positive gMG. The class includes monoclonal antibodies (eculizumab, ravulizumab, pozelimab), a macrocyclic peptide (zilucoplan), a small interfering RNA (cemdisiran) and a nanobody (gefurulimab). Two phase 3 trials, NIMBLE (cemdisiran) and PREVAIL (gefurulimab), were published in 2026. Previous pairwise and network meta-analyses have focused on efficacy and reached consistent conclusions. Their safety analyses were limited to crude totals: any adverse event, serious adverse events or treatment-related adverse events. Those totals do not address the concerns that drive clinical monitoring of C5 inhibitors: meningococcal and other serious infections treatment discontinuation injection-site reactions with subcutaneous agents long-term mortality Randomized periods last 12–26 weeks, which is too short to estimate rare harms. Open-label extensions add several hundred patient-years of exposure and allow exposure-adjusted rates of rare events. 2.2 Objectives Primary: to estimate the risk of serious adverse events, discontinuation due to adverse events, serious infections and death with C5-targeted monotherapy compared with placebo during randomized double-blind treatment. Secondary safety: to estimate the risks of other adverse-event categories, specific infections, injection-site reactions and disease-related harms, and to describe meningococcal infections. Long-term: to estimate exposure-adjusted rates of meningococcal infection and death, and the proportion of patients with a serious adverse event, in the open-label extensions. Secondary efficacy: to summarize the effect on MG-ADL, QMG and MGC scores, so that harms can be weighed against benefit. 3. Methods 3.1 Eligibility criteria Population: adults (18 years or older) with generalized myasthenia gravis of any antibody status. Excluded: ocular-only myasthenia gravis and paediatric-only trials. Intervention: any agent targeting complement C5 (antibody, peptide, siRNA, nanobody or other), given as monotherapy added to stable background therapy. Excluded: inhibitors acting only upstream of C5, such as factor B, factor D or C1q inhibitors. Comparator: placebo plus stable background therapy. Excluded: active comparators and uncontrolled studies. Outcomes: at least one extractable safety or efficacy outcome listed in section 3.6. Design: randomized, double-blind (or otherwise masked) placebo-controlled trials with parallel or crossover designs. Their open-label extensions are used for the long-term analysis. Excluded: non-randomized studies, single-arm studies, case series, reviews, and pharmacokinetic studies without clinical outcomes. Limits: none on language, date or publication status. Crossover trials contribute first-period data only. The trial is the unit of analysis: multiple reports of one trial are linked and counted once. 3.2 Arm selection in multi-arm trials Primary analysis: the arm receiving the regimen intended for clinical use, defined as the dose carried into phase 3 or the trial’s confirmatory monotherapy comparison (zilucoplan 0.3 mg/kg; cemdisiran 600 mg every 12 weeks). Other eligible monotherapy arms (other doses or agents, for example zilucoplan 0.1 mg/kg and pozelimab 200 mg every 4 weeks) enter sensitivity analyses. A trial’s designation of an arm as exploratory or confirmatory does not affect eligibility. Combination arms (two C5-targeted agents together) are outside the monotherapy scope. They enter one further sensitivity analysis that covers all C5-targeted regimens. Shared placebo groups are not counted twice. For binary outcomes, added arms are combined with the trial’s primary arm against the shared placebo group. For continuous outcomes, the placebo group is split equally between comparisons (Cochrane Handbook, section 23.3.4). 3.3 Information sources Bibliographic databases: PubMed/MEDLINE, Cochrane CENTRAL, Embase, Scopus and Web of Science Core Collection, each searched from inception. Trial registers and results databases: ClinicalTrials.gov, WHO ICTRP and the EU Clinical Trials Register. Regulatory documents: FDA labels and review packages, EMA assessment reports and Health Canada product monographs. Other sources: conference abstracts (AAN, AANEM, MGFA, EAN, MDA), sponsor posters, and the reference lists of included trials and relevant reviews. Search dates: the original searches ran to 29 July 2026. The update for long-term extension data ran to 30 October 2026. 3.4 Search strategy Three concept blocks are combined with AND: myasthenia gravis complement C5 inhibition, including each agent by name and development code a randomized-trial filter, applied to bibliographic databases only; no design filter is applied in trial registers A second search adds terms for open-label extensions and long-term safety. The full database-specific strategies, with dates and hit counts, are provided on OSF. PubMed example: ("Myasthenia Gravis"[Mesh] OR myasthen*[tiab]) AND ("Complement C5"[Mesh] OR "complement C5"[tiab] OR "terminal complement"[tiab] OR eculizumab[tiab] OR ravulizumab[tiab] OR zilucoplan[tiab] OR cemdisiran[tiab] OR pozelimab[tiab] OR gefurulimab[tiab] OR ALXN1210[tiab] OR ALXN1720[tiab] OR RA101495[tiab] OR ALN-CC5[tiab] OR REGN3918[tiab]) AND (randomized controlled trial[pt] OR randomi*[tiab] OR placebo[tiab] OR "open-label extension"[tiab] OR "long-term"[tiab]) 3.5 Study selection and data extraction Two reviewers independently screen titles and abstracts, then full texts. Disagreements are resolved by discussion and, if needed, by a third reviewer. Excluded full texts are listed with specific reasons, and the selection process is shown in a PRISMA 2020 flow diagram. Two reviewers extract data independently into a piloted form, and discrepancies are resolved against the source document. Each extracted value is recorded with its source (document, table or page) and a status code: Verified: read from the full-text article or its supplement. Exact, other source: printed in a registry results record, regulatory document, abstract or sponsor poster. Derived: back-calculated from a reported percentage and the safety-set denominator, and checked to reproduce the printed percentage. Provisional: from an ambiguous source, such as a sum of event terms where patients may overlap. Used in sensitivity analyses only. Sources are ranked as follows, highest first: peer-reviewed full text and supplements registry results regulatory documents conference abstracts and posters sponsor press releases Data items: Trial: registration number, phase, design, sites, randomization ratio, duration of the double-blind period, arms and doses. Population: inclusion criteria, antibody status, MGFA class, age, sex, disease duration, baseline MG-ADL and QMG scores, background therapy and refractory status. Analysis sets: safety set per arm (patients receiving at least one dose) and efficacy set (ITT or mITT). Outcome data per arm: number of patients with events for binary outcomes; adjusted mean differences with confidence intervals or standard errors