Cytokine-Induced Memory-like NK Cells Exhibit Massive Expansion and Long-Term Persistence after Infusion Post-Haploidentical Stem Cell Transplantation: A Report of the First Three Cases in a Phase I Trial

Background

Relapse of acute myeloid leukemia (AML) after allogeneic stem cell transplant has a poor prognosis with limited treatment options. Cytokine-induced memory like natural killer (CIML NK) cells are a novel therapy with enhanced cytotoxicity independent of KIR ligand interactions able to induce remission of relapsed/refractory AML (Romee et al, Science TM 2016). We are evaluating the safety and potential efficacy of donor-derived CIML NK cells in patients with relapsed myeloid malignancies after haploidentical donor transplant (haploSCT) in a phase 1/1b study (clinicaltrials.gov: NCT040247761). Here we report on manufacturing, safety, and in vivo correlative biology of the adoptively transferred CIML NK cells in the first 3 enrolled patients.

Methods

The primary endpoint is to identify the maximum-tolerated dose (MTD) of CIML NK cells in patients with MDS, MPN, or AML who relapsed after haplo-SCT. NK cells were enriched from donor-derived non-mobilized leukapheresis product via two-step CD3 depletion followed by selection of CD56+ cells using the CliniMACS reagent system (Miltenyi Biotec). The enriched NK cell product was cultured for 12-16 hours in X-VIVO 15 media containing rhIL-12, rhIL-15, and rhIL-18 to generate CIML NK cells (Figure 1A). Patients in the current cohort were lymphodepleted with fludarabine 25mg/m² daily for 3-5 days and cyclophosphamide 60mg/kg daily for 2 days followed by CIML NK cells at a dose of 5-10 x 10⁶ cells/kg and IL-2 10⁶ IU/m² QOD for 7 doses. Dose-limiting toxicities were evaluated for 6 weeks following NK cell infusion. Response to therapy was assessed at day+28 following CIML NK cell infusion.

Results

Patient #001 has FLT3-ITD AML that relapsed 5 months after a reduced intensity (RIC) haploSCT. She received 7.4x10⁶ NK cells/kg followed by IL-2 10⁶ IU/m² QOD for 7 doses. Her day+28 bone marrow had no leukemia blasts although FLT3-ITD mutation remained detectable. Two months post-CIML NK the leukocyte and granulocyte chimerism were 88% and 89%, respectively. Patient #002 has AML with multiple pathogenic variants, including a potentially pathogenic variant in TP53. His disease relapsed 15 months post haplo-SCT with repeat marrow showing all the original mutations, including the TP53 variant (VAF 50.5%). He received 9.5x10⁶ NK cells/kg followed by IL-2 10⁶ IU/m² QOD for 7 doses. His day+28 marrow had trilineage hematopoiesis without any mutations, including no TP53 mutation (Figure 1B). Two months post-CIML NK the leukocyte and granulocyte chimerism were both 99%. Patient #003 has MDS whose disease relapsed 8 months post RIC haploSCT with persistence of all her diagnostic pathogenic mutations. She received approximately 9.2x10⁶ NK cells/kg and has only just completed IL-2 10⁶ IU/m² QOD for 7 doses. Among the 3 patients, the main toxicity was prolonged cytopenia requiring stem cell boost in one case.

Donor NK cells demonstrated a dramatic shift from a predominantly CD56^dimCD16^hi (88% of NK cells) to a CD56^dimCD16^lo phenotype (49% of NK cells) as CIML NK cells. Infused CIML NK cells expanded massively, with approximately 50-fold, 10-fold, and 15-fold maximum in vivo expansion in the first three patients, respectively (Figure 1C). CIML NK cells were the major population in the day+28 marrows in both patients #001 and #002, with CD56+CD7+ cells constituting 89% of the cellularity in the former and 48% in the latter (Figure 1C). CIML NK cells persisted for 3 and 6 months post-infusion in the first two patients. The NK cells were predominantly mature, with most expressing CD16 and low levels of the inhibitory receptor NKG2A. PD-1 expression was much lower on the expanded NK cells vs the pre-infusion donor-derived NK cells. There was minimal concurrent expansion of CD4+CD25+ T-regulatory cells (Figure 1D).

Conclusion

We show with the first 3 patients in this trial that CIML NK cells can be generated and infused safely, can expand massively in the peripheral blood and bone marrow within the first 30 days post-infusion, and can persist for several months. In addition, CIML NK cell infusion can reduce the burden of pathogenic variant alleles to below the limit of detection, including the burden of high-risk mutations such as in TP53. Though our results are preliminary, the massive in vivo expansion and long-term persistence of adoptively transferred CIML NK cells underscores the unique biology of these cells that makes them an attractive option for cellular therapy protocols.