Descartes-08

Cartesian Therapeutics is testing Descartes-08 (resecabtagene autoleucel, also called rese-cel), an mRNA-engineered autologous CAR-T cell therapy targeting BCMA, in patients with generalized myasthenia gravis (MG, an autoimmune disease where antibodies block signals from nerve to muscle). The Phase 2b trial (NCT04146051, N=26 randomized: 15 Descartes-08 and 11 placebo) published in Nature Medicine in January 2026 showed 66.7% of Descartes-08 patients achieved a 5-point or greater improvement on the MG Composite scale at Month 3 versus 27.3% on placebo (p=0.0472), a mean MG-ADL change of -3.8 versus -1.7, with no cytokine release syndrome, no neurotoxicity, and no lymphodepleting chemotherapy required [1][2]. The confirmatory Phase 3 AURORA trial (NCT06799247) is now enrolling 128 patients with MG-ADL (the Activities of Daily Living scale, measuring chewing, swallowing, breathing, speaking, and lifting) as the primary endpoint, with topline readout guided to Q1 2027 and BLA planned mid-2027 [3][4]. This is the most clinically advanced CAR-T program in any autoimmune indication, putting Cartesian ahead of Kyverna (CD19-directed KYV-101, KYSA-6 Phase 3 first patient dosed December 2025, enrollment completion targeted mid-2027), Cabaletta, and larger competitors chasing the same B-cell depletion thesis [10]. For a single-asset small-cap trading under ticker RNAC, the Phase 3 result is close to a company-defining binary event [4]. Positive replication would put an autologous cell therapy on an autoimmune label for the first time and reset the pricing conversation for how transient mRNA-based CAR-T competes against permanent viral-vector CAR-T.

Descartes-08 is a novel investigational asset with no approvals in any indication. The Phase 2b trial in generalized MG (NCT04146051) randomized 26 patients (15 Descartes-08, 11 placebo, unequal 1.36:1 split at analysis after screen failures and withdrawals from an earlier ~36-patient allocation), is active but not recruiting, and reported statistically significant placebo-corrected improvement on the MG Composite scale at Month 3 as the pre-specified primary endpoint, with MG-ADL, QMG (Quantitative Myasthenia Gravis score, a physician-scored strength assessment), and MG-QoL15r (a 15-item quality-of-life questionnaire) as secondary endpoints [1]. Durability follow-up reported by Chahin et al. in Annals of Clinical and Translational Neurology in 2025 shows sustained response in the evaluable subset at 12 months, though attrition from the 26 randomized to 12 evaluable at Month 12 is material and warrants scrutiny [5]. The Phase 3 AURORA trial (NCT06799247, N=128) began enrolling in 2025 with MG-ADL as the primary endpoint, matching the FDA-preferred scale used for approvals of efgartigimod and ravulizumab [3]. Cartesian is running three additional Descartes-08 studies: Phase 2 in systemic lupus erythematosus (NCT06038474), Phase 2 in autoantibody-mediated myositis (NCT07391605), and Phase 1 in pediatric autoimmune disease (NCT07089121). The company holds RMAT designation (Regenerative Medicine Advanced Therapy, an FDA program giving priority interaction and support to cell and gene therapies with early evidence of major benefit) for the MG program [4]. Phase 3 topline is guided to Q1 2027, with BLA filing planned for mid-2027; no interim efficacy analysis has been publicly disclosed, so the readout is effectively binary [4][11].

BCMA (B-cell maturation antigen, gene TNFRSF17) is a receptor on the surface of plasma cells, the terminal antibody-producing cells of the immune system. Its natural ligands BAFF and APRIL keep plasma cells alive [6]. In generalized MG, long-lived plasma cells make autoantibodies against the acetylcholine receptor at the neuromuscular junction, blocking the signal from nerve to muscle and causing the fluctuating weakness that defines the disease. Kill the plasma cells and the antibody factory shuts down. Descartes-08 arms a patient's own T cells with a chimeric antigen receptor that recognizes BCMA and destroys any cell displaying it. The engineering trick that distinguishes Cartesian from every other CAR-T shop: the CAR is encoded on mRNA, not integrated into the genome by a lentivirus. mRNA degrades within days, so CAR expression is transient. That means no lymphodepleting chemotherapy is required before infusion, treatment is outpatient, and no cytokine release syndrome or ICANS (immune effector cell-associated neurotoxicity syndrome, a form of brain inflammation seen with conventional cancer CAR-T) has been reported at clinical doses [2][7]. The BCMA target itself is heavily validated: idecabtagene vicleucel (Abecma) and ciltacabtagene autoleucel (Carvykti) are two FDA-approved BCMA-directed CAR-T therapies for multiple myeloma, and rituximab (anti-CD20 B-cell depletion) already works in refractory MG. Descartes-08 goes one layer deeper than rituximab because mature plasma cells lose CD20 but keep BCMA. BCMA is therefore first-in-autoimmune, not first-in-class.

NCT04146051 was a randomized, double-blind, placebo-controlled Phase 2b in adults with generalized MG on standard-of-care immunosuppression. Twenty-six patients (15 Descartes-08, 11 placebo) received six weekly infusions of Descartes-08 or placebo without any lymphodepleting conditioning. Primary endpoint was the proportion of patients with an MG Composite improvement of at least 5 points at Month 3 (pre-specified in the protocol); secondary endpoints included MG-ADL (Activities of Daily Living), QMG (Quantitative Myasthenia Gravis physician-scored strength assessment), and MG-QoL15r (15-item quality-of-life questionnaire) [1]. Vu et al. reported 66.7% (10/15) of Descartes-08 patients hit the MGC 5-point threshold versus 27.3% (3/11) on placebo (p=0.0472), with mean MG-ADL change at Month 3 of -3.8 for Descartes-08 versus -1.7 for placebo (overall population), and Fedak et al. described accompanying reductions in circulating autoantibody titers and plasma-cell biomarkers [1][2]. Chahin et al. reported 33% of Descartes-08 patients achieved minimum symptom expression (MG-ADL score of 1 or less) by Month 6, sustained through Month 12 in the evaluable population [5]. The design was appropriate for Phase 2b: placebo-controlled, blinded, standard scales, active immunosuppression allowed on both arms. The N=26 sample size limits subgroup analysis, particularly across AChR-antibody-positive versus MuSK-antibody-positive patients. The Phase 3 AURORA trial (NCT06799247) scales enrollment to 128 patients and swaps to MG-ADL as the primary endpoint, matching the FDA-preferred scale used for argenx's efgartigimod and AstraZeneca's ravulizumab approvals [3]. Randomization ratio and any interim analysis structure are not disclosed in the public registry entry; the working assumption should be a single terminal readout in Q1 2027 [11].

Our model estimates a 4% chance this drug is eventually approved. It starts from the historical base rate for Phase 2 drugs in this area (about 30%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is held back by heavier-than-usual blinding, the sponsor's thin or weak approval record, weak or limited earlier-phase results, and smaller-than-typical enrollment for this phase. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Durability is the sharpest question. mRNA CAR-T is transient by design, and while Chahin et al. reported durable clinical benefit out to 12 months in the evaluable subset, only 12 of the original 26 randomized patients contributed to that Month-12 analysis, and the disclosed data on how long autoantibody suppression lasts and how repeat dosing performs is thinner than the single-dose acute data [5]. Efficacy risk on Phase 3 is real: MG placebo response is variable, and the Phase 3 must replicate the Phase 2b delta on MG-ADL specifically, which was a secondary endpoint in the Phase 2b (the primary was MGC). The placebo-corrected MG-ADL effect size (-2.1 points) is close to the commonly cited clinically meaningful threshold of 2 points, so a modest placebo-arm regression to the mean or a slightly weaker treatment effect could miss statistical significance in AURORA. Safety risk: BCMA CAR-Ts in myeloma have shown neurotoxicity, cytopenias, and infection risk from prolonged B-cell aplasia (loss of the B-lymphocyte compartment that produces antibodies, leaving patients vulnerable to infection). The mRNA transient design should mitigate these, but longer follow-up in the Phase 3 population will be the real test. Commercial risk is substantial. Vyvgart (argenx efgartigimod) generated $2.2B in total 2024 global net product sales across all approved indications (gMG plus CIDP, which was approved in the U.S. in mid-2023 and ramped strongly through 2024); the gMG-specific split is not separately disclosed but is the larger share, still well short of $2.2B in isolation [9]. Payers have no precedent for autoimmune CAR-T pricing, autologous manufacturing scale is expensive, and physicians accustomed to prescribing an injectable will need convincing that a cell therapy is worth the workflow. Off-treatment durable remission, not symptom improvement alone, is what would justify the switch.

Worth watching closely. Cartesian (RNAC) is a small-cap whose valuation is essentially a call option on Descartes-08 [4]. Cash position as of Q1 2026 was $120.4 million, extended by a $50 million K2HV term loan drawn in May 2026 and up to $100 million more available in tranches through 2028, taking runway into 2028 and past the Q1 2027 Phase 3 readout with room for the BLA cycle [11]. That runway removes the pure existential-cash tail risk and lets the story trade on scientific outcome rather than dilution mechanics. The Phase 3 AURORA MG-ADL readout is the binary event. Positive replication would put an autologous cell therapy on an autoimmune label for the first time, ahead of Kyverna (KYSA-6 Phase 3 of CD19-directed KYV-101 dosed first patient December 2025, enrollment completion targeted mid-2027, so their gMG readout trails AURORA by a year or more), Cabaletta, and the Bristol-Myers and Johnson & Johnson programs chasing similar B-cell and plasma-cell biology [10]. Check back at three milestones: Phase 3 enrollment completion (targeted 2026), interim safety updates, and topline readout (Q1 2027). Secondary catalysts are the SLE and myositis Phase 2 readouts, both currently guided to 2026 to 2027 windows in company communications, which will show whether the mRNA-CAR-T platform generalizes or whether MG was a lucky indication. The specific data point that would turn this from interesting to signal is a Phase 3 MG-ADL improvement of at least 3 points versus placebo with follow-up out to six months without a durability collapse. If Descartes-08 fails Phase 3, the mRNA-transient-CAR thesis for autoimmune weakens across the whole space and the field defaults back to permanent viral CAR-T strategies from Kyverna and the pharma majors.