Allele-specific targeting of CD33 as a novel approach to enhance CAR-T selectivity in relapsed acute myeloid leukemia post–allogeneic transplantation.

6554 Background: Relapse after allogeneic hematopoietic cell transplant (HCT) represents the most common treatment failure in acute myeloid leukemia (AML), urging the need for new therapies. Chimeric antigen receptor (CAR) T cells have shown efficacy in lymphoid and plasma cell malignancies but have not yet successfully translated to myeloid disease. This could in part be due to the expression of target antigens on normal myeloid cells, which might result in significant on-target but off-tumor effects. Exploiting a single nucleotide polymorphism (SNP) in SIGLEC3 , which encodes CD33, to target relapsing AML cells while selectively sparing donor-derived hematopoietic cells could improve the efficacy of anti-CD33 CAR-T cells in treating AML. Methods: We evaluated the in vitro efficacy and specificity of CAR-T cells that target each of two alleles of a common SNP in CD33 (rs2455069, CD33 Arg69Gly ). Controls included untransduced activated T-cells (UTD, negative control) and CAR-T cells targeting a different CD33 epitope (CART33 Pan positive control). Effector cells were incubated with CD33-expressing AML targets for 24 hours at serial effector-to-target ratios, and surviving target cells were quantified via flow cytometry to calculate percent cell killing. Initial screens used Jurkat cells lentivirally transduced with CD33 of each polymorphism (Jurkat33 Gly or Jurkat33 Arg ) to characterize CAR-T efficacy and specificity. Subsequent tests used MOLM-13 (homozygous CD33 Arg ) and KG-1 (homozygous CD33 Gly ) AML cell lines. Statistical comparisons of generated CAR-T cytotoxicity curves were performed with two-way ANOVA. Killing assays were performed with at least six biologic replicates. Results: CART33 Gly and CART33 Arg cells each showed statistically equivalent killing against their concordant Jurkat33 targets compared to CART33 Pan cells. Each CAR directed minimal killing of discordant Jurkat33 targets, similar to UTD controls. In KG-1 (CD33 Gly ) cells, CART33 Gly showed similar killing efficacy as CART33 Pan cells, while CART33 Arg were statistically equivalent to UTD cells at multiple E:T ratios. In MOLM-13 (CD33 Arg ) cells, CART33 Gly and CART33 Arg exhibited mixed efficacy and specificity: at lower E:T ratios (1:16 to 1:8), CART33 Gly was statistically equivalent to untransduced controls, but began to demonstrate more nonspecific toxicity at E:T above 1:4. CART33 Arg exhibited significant cytotoxicity once above E:T of 1:8, but only achieved similar toxicity to CART33 Pan at E:T ratios over 1:2. Conclusions: Allele-specific CART33 cells demonstrate promising efficacy and specificity in killing AML cells expressing either CD33 Gly or CD33 Gly . Further studies are underway to test efficacy in vivo against primary AML cells and safety toward normal hematopoietic progenitor cells that express the alternative CD33 allele.