Discovery

MD Anderson study identifies two mechanisms shaping allogeneic CAR T success

MD Anderson study identifies two mechanisms shaping allogeneic CAR T success

Researchers at The University of Texas MD Anderson Cancer Center have identified two distinct mechanisms governing whether allogeneic CAR T cells are rejected or expand productively in patients — findings with direct implications for the design of off-the-shelf cell therapies, according to a study published in Cancer Discovery.

Allogeneic CAR T therapies are manufactured in advance from healthy donors and can be deployed immediately, unlike autologous approaches requiring patient-specific manufacturing. A central obstacle has been inconsistent outcomes after infusion — cells are sometimes cleared by the host immune system before exerting therapeutic effect, and the determinants of rejection versus productive expansion have remained poorly understood.

The study, led by corresponding author Pavan Bachireddy, analyzed 11 patients with large B-cell lymphoma (LBCL) treated with a single manufacturing lot of cemacabtagene ansegedleucel (cema-cel), an allogeneic anti-CD19 CAR T product under development by Allogene Therapeutics. Using a single lot across all patients reduced product-level variability and helped isolate recipient- and cell-level determinants.

On the host side, patients who failed to expand cema-cel had high pre-existing frequencies of alloreactive CD8+ T cells that rapidly cleared the infused product. On the product side, among patients where rejection did not occur, robust clonal expansion was driven by effector-like features within the cells — not by stem or central memory phenotypes, which had been assumed to be optimal for persistence.

The expansion-associated findings were reproduced in two independent cohorts: one receiving a separate cema-cel manufacturing lot, and one receiving Allogene's allogeneic anti-BCMA CAR T candidate ALLO-715 from a completed Phase I study (NCT04093596). The replication across a different product lot and a distinct construct targeting a different antigen strengthens the generalizability of the findings.

The AllSci BriefFree, systematic R&D and deal news. Daily.

Cema-cel, developed by Allogene Therapeutics (Nasdaq: ALLO), is now the subject of a pivotal, randomized Phase II trial (NCT06500273) evaluating first-line consolidation in minimal residual disease (MRD)-positive large B-cell lymphoma (LBCL) patients after standard chemoimmunotherapy, and holds both FDA Regenerative Medicine Advanced Therapy (RMAT) and Fast Track designations; an interim futility analysis reported in April 2026 showed 58.3% MRD clearance in the cema-cel arm versus 16.7% in the observation arm, and the trial continues to enroll.

ALLO-647, a companion lymphodepletion antibody, was permanently discontinued in August 2025 following a Grade 5 adverse event — hepatic failure secondary to disseminated adenovirus infection — and cema-cel now proceeds with standard fludarabine and cyclophosphamide lymphodepletion only.

The competitive field has converged on several engineering strategies to address host rejection. San Francisco-based Caribou Biosciences' CB-011, an allogeneic anti-BCMA CAR T in Phase I for multiple myeloma, uses HLA class I disruption combined with HLA-E expression to evade both host T cell and natural killer cell killing. St. Louis-based Wugen's soficabtagene geleucel (WU-CART-007), an allogeneic anti-CD7 CAR T in pivotal Phase II for relapsed or refractory T-cell acute lymphoblastic leukemia, received FDA Breakthrough Therapy Designation in January 2026. The mechanistic framework from the MD Anderson paper — identifying pre-existing alloreactive CD8+ T cell frequency as a rejection determinant and effector-like phenotype as a driver of expansion — suggest potential biomarkers for identifying patients at risk of rejecting donor-derived CAR T cells and may inform donor selection, manufacturing, and engineering strategies for future allogeneic products


Access the AllSci platform to explore the science behind the news.


Spot something wrong? Report an issue with this article