A 4-fold difference in complete remission rate — 50% versus 12.5% — depending solely on which donor’s NK cells were used to manufacture SENTI-202 is not a minor process variable. It is the central fact reshaping the entire registrational strategy for this Logic Gated CAR-NK therapy in relapsed/refractory AML, and it arrived not from a prospectively designed biomarker study but from a post-hoc covariate analysis of 22 patients.
The “Donor X” phenotype, deliberately anonymized in the release, is present in roughly half of adult donors and drives increased NK cell cytotoxicity through a mechanism independent of HLA or KIR matching — which preserves the allogeneic, off-the-shelf manufacturing model. Retroactive preclinical confirmation in an MV4-11 NSG mouse model showed median survival extended beyond the observation window for Donor X-derived SENTI-202, compared to a finite median survival for non-Donor X product. That preclinical corroboration matters: it transforms a statistical artifact into a biologically plausible signal. Senti will now restrict all future manufacturing — including pivotal trial supply — exclusively to Donor X-derived cells, effectively redefining the product itself mid-program.
The FDA Type B RMAT meeting endorsed a single-arm, multi-center registrational design, which is the appropriate architecture for a disease where historical control rates are well-documented and durable MRD-negative remissions are rare. At the RP2D across the full 22-patient cohort — a mix of Donor X and non-Donor X material — 100% of complete remissions were MRD-negative, and every patient who had achieved CR or CRh as of the 2025 ASH data cut remains in remission with an additional seven months of follow-up, the longest now exceeding 21 months. That durability in R/R AML, a setting where median overall survival is typically measured in months, is the clinical anchor for the entire regulatory argument.
The pivotal trial’s success now hinges on one manufacturing question that the Phase 1 data cannot fully answer: whether restricting exclusively to Donor X donors introduces supply constraints or lot-to-lot variability at commercial scale. The single number to track is the cCR rate in the first pre-specified interim analysis of the registrational trial — if Donor X selection genuinely eliminates the 12.5% responder pool from contaminating the efficacy signal, the trial’s statistical power assumptions become considerably more defensible than the blended Phase 1 data alone would suggest.
Jon Napitupulu is Director of Media Relations at The Clinical Trial Vanguard. Jon, a computer data scientist, focuses on the latest clinical trial industry news and trends.
