A Phase I Study of Intravenous NCI IL-15 to Enhance Adoptively Transferred NK Cells Uncovers Defects in CD16 Triggered IFNγ Production and Competition Between Donor NK and Recipient T Cells

Adoptive transfer of IL-2 driven NK cells can induce remissions in patients with refractory AML but this may be limited by IL-2 induced expansion of regulatory T cells (Tregs). Because IL-15 does not stimulate Tregs, we performed a phase I dose escalation trial of recombinant human IL-15 (NCI Frederick) to enhance adoptive NK cell transfer.  Patients with refractory AML received a lymphodepleting regimen of cyclophosphamide and fludarabine, followed by infusion of haploidentical NK cells activated overnight with IL-15 and intravenous IL-15 for 12 daily doses in cohorts of 0.25, 0.5, 1 and 0.75 mcg/kg.  We treated 24 patients and deterimined the maximum tolerated dose (MTD) to be 0.75 mcg/kg. IL-15 serum levels peaked 1 hour after dosing with a significant drop at 4 hours and no evidence of drug accumulation between doses 1 and 6 (Figure, A). Fevers were common, possibly attributable to the rise in multiple inflammatory cytokines 4 hours after IL-15 dosing (Figure, B). Higher fevers occurred after 6 doses, corresponding to higher cytokine levels. No patients experienced hypotension or vascular leak. Dose limiting toxicity was observed in 2 subjects at 1.0 mcg/kg who did not recover neutrophils 42 days after NK cell infusion, one of whom died of infection 60 days after adoptive NK cell transfer at which time donor NK cells persisted in blood and pleural fluid. Clearance of refractory leukemia was seen in 38% of patients 14 days after adoptive transfer. Three patients with CRp went on to best donor allogeneic transplant with overall survival of those receiving IL-15 (n=24) of 32% at 12 months. We observed two patterns of donor cell expansion and autologous count recovery. In the 5 of 24 (21%) patients with detectable NK cell donor chimerism, the mean ALC measured at day 7 after NK infusion was 160/µL with an mean of 56% donor chimerism. By day 14 the mean ALC increased to 1200/µL with 87% donor chimerism, where virtually all donor lymphocytes were NK cells. The remaining 19 patients who did not expand NK cells had a mean ALC of 68/µL with 27% donor chimerism at day 7, which increased to 900/µL and only 2.5% donor chimerism by day 14. In the non-NK expanders, nearly all of the day 14 lymphocytes were recipient CD8⁺ T cells, suggesting host T cell mediated rejection of donor NK cells. We extensively evaluated the function of the successfully in vivo expanded donor NK cells collected from the recipient day 14 post-NK infusion and IL-15 administration, and directly compared it to that of NK cells at steady state from the same donor. Although we expected degranulation to be increased in the IL-15 expanded NK cells, CD107a expression was similar in steady state donor and in vivo IL-15 expanded NK cells (Figure, C). In contrast, IFNγ production triggered by CD16 signalling with Rituxan against antibody-dependent cellular cytotoxicity sensitive Raji targets was significantly decreased in the in vivo IL-15 expanded NK cells (Figure, D). This did not reflect an inherent defect in IFNγ production as IL-12/IL-18 stimulation of day 14 in vivo IL-15 expanded donor NK cells induced significantly more IFNγ compared to the study state donor cells (Figure, E). These characteristics (decreased target cell-induced INFg and increased IL-12/IL-18 induced IFNγ) suggest that donor NK cells expanded in vivo with IL-15 are functionally similar to the CD56^bright population.  This effect is not unique to IL-15 as a similar functional pattern was observed in patients with in vivo IL-2 expanded NK cells. This suggests that robust, cytokine driven in vivo NK proliferation may limit the target cell-induced cytokine production of the expanded population. In summary, NCI IL-15 led to robust donor NK cell expansion in some patients, but competitive stimulation of recipient CD8⁺ T cells was common. Even when in vivo expansion of donor NK cell was successful, the cells exhibited defects in target cell and CD16-induced IFNγ production. Long-lived CMV-driven adaptive NK cells with memory like properties are associated with enhanced anti-leukemia activity and CD16 signalling. Novel methods to expand adaptive NK cells may be of value for the next generation of NK cell therapy. Additionally, the NCI CITN trial of subcutaneous IL-15 administration suggests that compared to IV dosing, it may favor donor NK vs. host T cell expansion based on different affinities of trans-presented IL-15 to these cell populations, and these strategies should be tested to optimize NK cell adoptive transfer.