The optimized CD33-directed CAR T-cell therapies FL33-02 and FL33-03 generated complete remissions in patients with relapsed/refractory acute myeloid leukemia (AML), according to data from a phase 1/2 trial (NCT06326021) presented in a poster session at the
All patients treated with FL33-02 (n = 6) achieved a complete remission with incomplete hematologic recovery (CRi). Among patients treated with FL33-03 (n = 5), 60% achieved a CRi, 20% had a partial response, and 20% did not respond. At a data cutoff of May 15, 2026, the median follow-up was 2.5 months (range, 0.6-17.8). The primary end point of the trial was safety, with efficacy assessed as a secondary end point.
“[The] FL33-03 structure [has] potent antitumor effects while also having the function of downregulating TNFα and improving safety,” lead study author Jing Pan, MD, of Beijing GoBroad Hospital in China, and colleagues wrote in a poster presentation of the data. “However, infection remains a prominent safety concern; early bridging hematopoietic stem cell transplant may reduce infection-related risks.”
The multicenter, single-arm, open-label phase 1/2 trial evaluated optimized CD33-directed (FL33) CAR T cells in patients with post-transplant and transplant-naive relapsed/refractory AML. The constructs were developed to address the safety limitations of a prior optimized CD33 CAR T-cell therapy (FL33-01), which produced antileukemic activity but was associated with grade 4 cytokine release syndrome (CRS) linked to elevated serum IL-6, IFN-γ, and TNF-α that was unresponsive to tocilizumab (Actemra) or corticosteroids.2 FL33-03 incorporates further structural optimizations of FL33-02 intended to enhance safety.
Patients treated with FL33-02 were assigned to 3 cohorts by CAR T-cell source: autologous (cohort A), pretransplant donor–derived (cohort B), or new donor–derived (cohort C). Using a BOIN12 design, phase 1 identified the optimal biological dose via dose escalation across 2 dose levels (5.0 × 10⁵ and 1.0 × 10⁶ cells/kg), followed by phase 2 expansion. Cohorts A and B received standard lymphodepletion, and cohort C received advanced lymphodepletion with cyclophosphamide 30 mg/kg for 3 days. Patients treated with FL33-03 were stratified into 2 groups based on the administration of TNF-α inhibitors, with the objective for that construct restricted to safety evaluation. HSCT was guided by treatment response, remission, and recurrence.
From July 2024 to May 2026, 11 patients were enrolled; the first 6 received FL33-02 (cohort A, n = 2; cohort B, n = 1; cohort C, n = 3) and the remaining 5 received FL33-03 (group 1, n = 3; group 2, n = 2). All patients except those in FL33-03 group 2 received concomitant TNF-α inhibitors.
Across the population (n = 11), the median age was 16 years (range, 7-46), and 63% of patients were male. Patients had received a median of 5 prior lines of therapy (range, 3-9); 45% had undergone prior allogeneic HSCT, and 37% had received prior CAR T-cell therapy. The median bone marrow blast count by flow cytometry was 38.9% (range, 0.4-88.5), and 27% of patients had extramedullary disease.
CAR T-cell expansion was robust with both constructs. Among patients treated with FL33-02, the median peak transgene level was 3690.5 copies/μg (range, 817-171,145). Expansion was more pronounced with FL33-03, which was associated with a median peak transgene level of 12,288 copies/μg (range, 1490-391,583).
CRS occurred in all 11 patients and was grade 1 or 2 in 9 patients (82%) and grade 3 or 4 in 2 patients (18%); one patient in the FL33-02 group developed grade 3 CRS, and the remaining patients experienced grade 1 or 2 CRS. All patients experienced grade 3/4 cytopenias. Immune effector cell–associated neurotoxicity syndrome occurred in 40% (2 of 5) of patients treated with FL33-03, and macrophage activation syndrome/hemophagocytic lymphohistiocytosis (MAS/HLH) occurred in 33.3% (2 of 6) of patients treated with FL33-02. Two of 11 patients developed GVHD. Investigators reported that grade 3 or higher infections occurred in 3 patients within 30 days of infusion.
Serum TNF-α levels tracked with the construct received. Within 1 month of infusion, all 6 patients treated with FL33-02 had a marked increase in serum TNF-α, whereas 3 of the 5 patients who received FL33-03 experienced only a mild increase, consistent with the construct's intended TNF-α–modulating effect.
