Putative Predictors of Response to WU-NK-101, an Allogeneic, Enhanced Memory (ML) Natural Killer (NK) Cell Therapy Product, for Relapsed/Refractory (R/R) Acute Myeloid Leukemia (AML)

Strategies to enhance NK-cell activity prior to adoptive transfer into patients with AML have yielded inconsistent clinical benefit. Immune cells in the AML tumor microenvironment (TME) are dysfunctional, in part due to enhanced interferon-γ signaling and acquisition of a senescent-like T/NK phenotype (Vadakekolathu, et al. Sci. Transl. Med. 2020). Cytokine-induced memory-like (CIML) NK cells are more potent for anti-tumor responses than conventional NK (cNK) cells, are well tolerated, and active in patients with R/R AML (NCT#01898793) with complete response (CR)/CR with partial/incomplete hematologic recovery rate of 47% and a median DOR of 9.4 months after a single infusion (Berrien-Elliott, et al. Cancer Discov. 2020). WU-NK-101 is a cell therapy derived from healthy donor NK cells, which are activated to induce a ML-NK cell program, expanded, and frozen to create an off-the-shelf cell therapy. This abstract 1) describes the approach for treatment of R/R AML with WU-NK-101, built upon prior clinical experience with CIML-NKs; 2) characterizes TME gene expression (GE) signatures associated with outcome after the infusion of CIML-NK in R/R AML and 3) functionally characterizes WU-NK-101.

Bone marrow (BM) aspirates and trephine biopsies for immune GE profiling and spatially resolved proteomics (IO360® panel, n = 740 genes, and GeoMx® DSP; NanoString Technologies) were collected at baseline (BL) and on treatment (OT) from 15 patients receiving CIML-NK cells on study NCT #01898793. ssGSEA-based scores were computed using the IOBR package in R. Machine learning was used to select features that correlated with response. Cell phenotypes were evaluated by mass cytometry. NK cell function was assayed in either conventional (N-; 20% IMDM) or TME-aligned media. Multi-omics studies were performed by mass spectrometry. Cell trafficking to TME was measured in NSG mouse tumors by tracking fluorescent-labeled WU-NK-101.

Correlative studies showed that BL GE signatures reflecting heightened neutrophil infiltration, chemokine receptor expression, and cancer antigen presentation predicted response to CIML-NKs (AUROC = 0.875, 0.875 and 0.850, respectively). We computed an 8-gene signature score (RPTOR, IRF9, TLR1, NLRP3, CD7, IL6R, FUT4, IL1RN) that was significantly higher at BL in responders and that correlated with response duration (Fig. 1A). Immune deconvolution of GE data with CIBERSORT highlighted critical differences in TME cellular composition at BL in responders to CIML-NKs, including a higher inferred abundance of macrophages, gamma/delta T cells and activated dendritic cells (Fig. 1B). In OT samples from responders, glycolytic GE programs as well as purine/pyrimidine and amino acid (AA) metabolism were upregulated compared with BL; no TME modulation was observed in non-responders (Fig. 1C). Analysis of WU-NK-101 cells revealed a unique phenotype compared with cNK cells, including higher expression of activating receptors, CD25, CD69, and GZMB, which positions WU-NK-101 for rapid activation and improved cytotoxicity. The latter hypothesis was confirmed by in vitro assays, with higher half maximal effective concentration (mEC50) for cNK vs WU-NK-101 (mEC50 = 5.2 vs 1.7; P = 0.032; Fig. 1D). The adverse impact of immunosuppressive TME-aligned media on cytotoxicity was evident on cNK but negligible to WU-NK-101 (mEC50 = 16.3 vs 2.0; P = 0.018; Fig. 1D). In N-media, WU-NK-101 used glucose as its main nutritional source; in hypoglycemic TME-aligned media, AA metabolic pathways were upregulated, indicating metabolic adaptability. This was supported by increased cell surface expression of AA transporters. WU-NK-101’s homing to the BM averaged 16.5 ± 2.4%, which is 5.5-fold higher than previously reported for feeder-cell+IL-2-expanded NK cells (~3%; Sato et al. Clin. Cancer Res. 2020).

In summary, an 8-gene TME signature showed excellent predictive ability for response to ML-NKs. In responders, ML-NK infusion was associated with TME modulation and with metabolic re-programming. Compared to cNK cells, WU-NK-101 had enhanced trafficking to the BM, anti-tumor activity, and a metabolic profile consistent with aerobic glycolysis, potentially contributing to mitigating TME adversity. Due to their resilience to an immunosuppressive TME, WU-NK-101 cells may represent an effective treatment modality for R/R AML; a phase 1 study of WU-NK-101 in R/R AML is in development (NCT #05470140).