In Vivo CAR-T in Lupus and Scleroderma: 14 Patients, no Lymphodepletion Save

Three reports at ACR Convergence 2026 describe CD19 CAR-T cells generated inside the patient, with no apheresis and no ex vivo manufacturing step. None of the three used fludarabine and cyclophosphamide conditioning.

The most striking of them changes the route of administration. A group from Ruijin Hospital (Abstract 0877) dosed six patients with refractory SLE using subcutaneous RXIM002, a CD7-targeted lipid nanoparticle carrying circular RNA. Two received 0.03 mg/kg and four received 0.06 mg/kg. CAR+ T cells appeared within 24 hours and persisted 18 to 60 days, and a lymph node sampled at day 18 showed 13.5% CAR+ T cells, enriched in CD8+ effector memory and TEMRA subsets. Mean time to at least 90% peripheral B cell depletion was 5.3 days with a median of 1 day, and two patients had B cell aplasia beyond two months. Mean SLEDAI-2K fell 8.3 points, complement normalized, anti-dsDNA and ENA titers fell, and all six discontinued immunosuppressants. CRS was grade 1 only, with no ICANS and no infections reported.

Speed of depletion is the easy part for a lipid nanoparticle platform. The harder question is whether a transient CAR reaches tissue, and Abstract 0263 addresses it directly. Five patients with refractory SLE received HN2301, a CD8-targeted LNP carrying CD19 CAR mRNA, intravenously on a repeated schedule. CAR T cells peaked at six hours and returned to baseline within 48 to 72 hours. The group took biopsies. Peripheral depletion was complete in 5 of 5, bone marrow depletion in all 3 patients evaluated, and lymph node depletion in 3 of 5. Peripheral aplasia lasted two weeks to three months. Anti-dsDNA and anti-nucleosome antibodies normalized or dropped, and C3 and C4 rose in every patient. All five reached SRI-4 by month 1, and two met LLDAS and DORIS with SLEDAI-2K of 2 to 4 at last follow-up. A single grade 1 ICANS resolved within 24 hours, and no grade 3 or higher CRS occurred.

The third report takes the approach out of lupus. Investigators at Wuhan No. 1 Hospital (Abstract 0223) treated three women with refractory diffuse systemic sclerosis using a single intravenous dose of JY231, a lentiviral vector that transduces T cells in vivo. Ages ranged from 29 to 36, median disease duration was 44 months, and all three had interstitial lung disease. CAR T cells expanded in all three, peaking between day 10 and day 17 and reaching 2% to 62% of CD3+ cells. B cells were depleted by day 9 to 10 and began reconstituting at month 2. mRSS, EUSTAR-AI and HAQ-DI improved and anti-Scl-70 declined. CRS was grade 1 in two patients and grade 2 in one, with onset on day 7 and resolution by day 8. There was no ICANS. One serious adverse event, abnormal liver function at day 61, was attributed to CMV.

Fourteen patients across three platforms, with follow-up measured in months. Added to the reports presented at EULAR, that is close to the whole clinical experience with in vivo CAR-T in rheumatology to date. Measured against autologous CD19 products at two and three years the efficacy evidence remains thin, a 5.3-day depletion curve is not an immune reset, and Abstract 0263 is the only one of the three that looked inside tissue.

The logistics are the finding. Apheresis, a manufacturing slot and fludarabine-cyclophosphamide conditioning are what keep CAR-T inside a small number of academic centers, and all three reports removed those steps and still depleted B cells. One did it with an injection under the skin and grade 1 CRS. For in vivo platforms the delivery case is well ahead of the efficacy case, and at this stage that is the right order. Autologous CAR-T has already shown what deep CD19 depletion does in these diseases. What these three programs test is whether the same biology survives a simpler delivery, and the number worth following over the next year is lymph node depletion.