BCMA-directed mRNA CAR T cell therapy for myasthenia gravis: a randomized, double-blind, placebo-controlled phase 2b trial.

The paper cites prior research establishing that MG is driven by autoantibodies from BCMA-positive plasma cells, discusses limitations of current treatments, and cites a prior open-label study of Descartes-08 to justify the trial. Strengths and weaknesses of prior work are acknowledged (e.g., limitations of conventional CAR T in autoimmune diseases, potential for infusion reactions). The rationale linking BCMA targeting to clinical improvement is logical.

Randomization was computer-generated with a permuted-block scheme, centrally assigned, and concealed. Blinding was maintained with opaque covers for infusion sets. A power analysis was performed (80% power to detect 47% difference), but only 26 (15/11) instead of the planned 30 participants comprised the primary efficacy population, reducing power. The primary endpoint was changed from MG-ADL to MGC at month 3 based on task force recommendations, which was disclosed but introduces a risk of bias. Inclusion/exclusion criteria were clearly defined. Pre-specified analysis was mITT. Outlier handling not explicitly described. No independent replication cohort, but this is understandable for a phase 2b trial. Sub-criteria: randomization_method adequate, randomization_unit adequate, blinding_levels adequate, power_analysis reported_but_inadequate (due to unmet sample size and endpoint change), inclusion_exclusion adequate, outlier_handling not_reported, controls (placebo) adequate, replicate_distinction and independent_replication not_applicable for a human RCT.

Sex is reported (66.7% female in Descartes-08, 54.5% in placebo) and is mixed so justification for single-sex not needed. Age and weight are reported as means. Comorbidities are not listed individually but health status via inclusion criteria is described. Species/strain and housing not applicable for a human study. Demographics table includes age, sex, ethnicity, disease duration, antibody status. Sub-criteria: sex_reported adequate, sex_justified not_applicable, age_weight_health adequate, demographics adequate, species_strain_source not_applicable, housing_conditions not_applicable.

The trial was approved by central and local IRBs (WCG IRB, ethics committees at University of Toronto, University of Alberta, Istanbul University) and a named regulatory authority. Written informed consent was obtained from all participants. Compliance with the Declaration of Helsinki and ICH E6 GCP guidelines is stated. Sub-criteria: irb_ethics_statement adequate, informed_consent adequate, regulatory_compliance adequate, iacuc_statement not_applicable.

The investigational product is autologous BCMA-directed mRNA CAR T cells, manufactured under GMP; dose is given as 52.5 × 10^6 viable CAR+ cells per kg ±45%. Placebo is 'acellular placebo lot' matched in appearance. Cytokine analysis used S-PLEX, V-PLEX panels from Meso Scale Discovery. Statistical software: SAS version 9.2 or higher, Mathematica version 11.0 or higher. No antibodies, cell lines, or organisms are used in this human clinical trial; those criteria are n/a. Mycoplasma testing is n/a. Reagents_identified is reported adequately for the investigational product and cytokine panels.

Primary endpoint uses a two-sample proportion test (Wald chi-squared); test is named. Assumptions of the test (independence, expected frequencies) are plausible but no verification of model fit is reported. Exact p-values: P = 0.0472 and P = 0.0258 are reported. Effect sizes: difference in proportions with 95% CI is reported for the primary outcome (0.39, 95% CI 0.01–0.77). Software identified (SAS, Mathematica). Data presentation: figures show means with 95% CI, individual data points are not shown for small n (n=15,11) which would be desirable. Mathematical plausibility: no obvious errors in the primary analysis, but some percentages in subgroups have small denominators where rounding could be checked but no arithmetic impossibility is apparent. Sub-criteria: tests_named adequate, assumptions_verified not_reported, exact_p_values reported_but_inadequate (only two p-values, most other comparisons not formally tested as secondary), effect_sizes_ci adequate for primary, software_identified adequate, data_presentation reported_but_inadequate (no individual points, bar-only), mathematical_plausibility pass.

The data availability statement says: 'Access to anonymized trial-level data (analysis datasets) and/or the study protocol will be provided upon request to qualified researchers conducting independent, rigorous research, after the review and approval of a research proposal and statistical analysis plan, as well as the execution of a data sharing agreement.' This meets the minimal threshold but is vague (no platform, no timeline beyond '12 months'). No repository deposit or accession numbers are provided; the statement is managed access for patient-level data, which is standard but not fully adequate per the scoring criteria. Code sharing is not mentioned.

Methods detail the trial design, randomization, blinding, endpoints, and statistical analysis. Trial registration number is provided (NCT04146051). A CONSORT flow diagram (Figure 1) is provided; the paper references a 'Reporting Summary' linked to the online content, which fulfills the reporting guideline requirement. All pre-specified outcomes appear reported (primary and secondary endpoints, safety, post hoc analyses). Limitations are discussed in detail (small sample, imbalances, seronegative overrepresentation, potential unblinding due to IRRs, etc.). Conclusions are proportional: the paper uses cautious language (e.g., 'may be achievable', 'preliminary data') and does not overclaim. Funding sources (Cartesian Therapeutics, NIH grants) and competing interests are fully disclosed.