← Back
NeuroTrials.ai
Neurology Clinical Trial Database

Descartes-08

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

Year of Publication: 2026

Authors: Tuan Vu, Hacer Durmus, Michael Rivner, ..., on behalf of the MG-001 Study Team

Journal: Nature Medicine


Clinical Question

Does Descartes-08, an autologous BCMA-directed mRNA CAR T cell therapy administered as six once-weekly outpatient infusions without preconditioning chemotherapy, improve clinical outcomes compared to placebo in patients with generalized myasthenia gravis?

Bottom Line

Six once-weekly infusions of Descartes-08 significantly improved MG composite scores at 3 months compared to placebo in patients with gMG, with 66.7% of treated patients achieving a meaningful clinical response. Responses deepened over time, were sustained through 12 months, and a third of all patients achieved minimum symptom expression. The therapy was well tolerated without lymphodepletion, B cell depletion, or hypogammaglobulinemia.

Major Points

  • Phase 2b randomized double-blind placebo-controlled trial (MG-001 part 3); NCT04146051
  • Descartes-08 uses mRNA rather than viral vector to encode the CAR, resulting in transient expression without need for preconditioning chemotherapy, enabling outpatient administration
  • Primary endpoint met: MGC responders (>=5-point reduction) at month 3: 66.7% (10/15) vs. 27.3% (3/11); p = 0.0472
  • AChR autoantibody-positive subgroup: 63.6% vs. 12.5% MGC responders at month 3 (p = 0.0258)
  • Responses deepened over 12 months: mean MG-ADL change -5.5 at month 4, maintained at -4.8 at month 12; QMG reached maximum improvement of -6.0 at month 12
  • 83% of treated patients achieved sustained clinically meaningful response at month 12
  • Minimum symptom expression (MSE; MG-ADL <=1) achieved by 33.3% of Descartes-08 patients by month 6, all sustained through month 12
  • Biologic-naive patients showed superior response: 88.9% MGC responders at month 3 and 55.6% achieving MSE by month 6, maintained through month 12
  • Median prednisone equivalent dose decreased by 55% in the Descartes-08 group at 12 months (20 mg to 9 mg); no participant required rescue escalation
  • Primary outcome (MG-ADL) was changed mid-study to MGC before database lock, based on MGFA Task Force recommendations; this is a notable methodologic concern
  • No grade 4 or 5 adverse events; no B cell depletion, no hypogammaglobulinemia, no cases of cytokine release syndrome meeting formal criteria
  • Infusion-related reactions (IRRs) were the most common AE: 80% in Descartes-08 vs. 56.3% in placebo; mostly grade 1-2, managed outpatient
  • HSV reactivation occurred in 3 Descartes-08 patients (15%), all with prior labial herpes on concomitant azathioprine
  • Notable placebo response, particularly in triple seronegative participants; a transient leukapheresis effect could not be excluded
  • Confirmatory phase 3 trial (AURORA, NCT06799247) enrolling AChR-positive patients

Design

Study Type: Phase 2b randomized double-blind placebo-controlled trial

Sample Size: 36


Criticisms

  • Primary endpoint was changed mid-study (before database lock) from MG-ADL to MGC score, raising concerns about post-hoc endpoint modification even if pre-specified in amendment v3.3
  • Small sample size (n=26 in mITT) with only 80% power to detect the primary outcome, making secondary outcome comparisons purely descriptive and precluding formal statistical testing
  • Efficacy population restricted to academic center participants, excluding community clinic enrollees, which limits generalizability and introduced unequal allocation between arms
  • Lack of randomization stratification by site contributed to imbalances in baseline characteristics (disease duration longer and thymoma rate higher in placebo group)
  • Notable imbalances in concomitant medications: more nonsteroidal immunosuppressants in placebo, more complement inhibitors and corticosteroids in Descartes-08 group
  • Significant placebo response, particularly in triple seronegative participants (overrepresented at 23.1%), limiting interpretation of treatment effect
  • Transient immunosuppressive effect of leukapheresis on the placebo group cannot be excluded as a contributor to placebo response
  • Protocol prohibited changes in nonsteroidal immunosuppressants throughout the study, potentially limiting the observed steroid-sparing effect and preventing drug-free remission
  • IRRs (fever, chills, myalgias) were more common in the Descartes-08 group and may have introduced unblinding bias in the assessment of MG severity scales
  • Absence of crossover data (placebo to Descartes-08) in the current report limits comparison of outcomes
  • AChR autoantibody testing by radioimmunoassay may be insufficiently sensitive to detect changes in antibody subtype or epitope specificity
  • Sponsor (Cartesian Therapeutics) was involved in study design, execution, and analysis; multiple authors hold equity in Cartesian Therapeutics

Funding

Cartesian Therapeutics (study product, design, execution, analysis); National Institute of Neurological Disorders and Stroke/NIH grants 1R44NS115426-01A1 and 1R44NS137943-01

Based on: Descartes-08 (Nature Medicine, 2026)

Authors: Tuan Vu, Hacer Durmus, Michael Rivner, ..., on behalf of the MG-001 Study Team

Content summarized and formatted by NeuroTrials.ai.