Expansion of Donor NK Cells for Adoptive Immunotherapy in Haploidentical Stem Cell Transplantation: A Phase I–II Study

Natural killer (NK) lymphocytes are essential for anti-cancer defense. Transplantations of haploidentical hematopoietic stem cells (HSCT) across HLA class I barriers highlighted the graft-versus-leukemia effect of alloreactive NK cells and strengthened prospects for NK cell immunotherapy in human cancer. Reconstituting NK cells remain immature with impaired cytotoxicity for more than 6 months after HSCT, suggesting a benefit of adoptive transfer of mature NK cells to provide the patient with competent cytotoxic effectors. We previously reported the feasibility of purifying NK cells in numbers sufficient for infusions of 1×107 cells/kg (Passweg et al. 2004). Here we describe an approach of increasing the effector to target ratio by large-scale good manufacturing practice (GMP)-compliant expansion of NK cells and preemptive multiple donor lymphocyte infusions (NK-DLI) after haploidentical HSCT for hematological malignancies. This single centre phase I–II clinical study was approved by the ethical committee. NK cells were purified from a median of 7.4L (range: 5.6–10.0L) unstimulated leukapheresis of 5 healthy donors by T cell depletion and subsequent NK cell selection with anti-CD3 and anti-CD56 coated immunomagnetic microbeads on the CliniMACS_ device. A median of 5.32×108 (2.87–6.42×108) CD56+CD3- NK cells was obtained with a purity of 98.5% (71.9–99.6%), and a residual CD3+ T cell content of 0.030% (<0.001–0.070%) corresponding to a T cell depletion efficiency of 3.51 log (3.15–4.87 log). NK cells were cultured in 30 air-permeable bags and up to 7.8L of GMP-certified medium containing human serum, IL-2, anti-CD3 monoclonal antibody OKT-3 and irradiated autologous feeder cells. After 15–20 days of culture, NK cell numbers increased on average 50 fold (12–137 fold). CD3+ T cells concomitantly expanded to 0.59% (0.014–1.68%) of total cells. A second CD3 depletion was performed with 3 NK cell products which exceeded the maximal T cell dose of 0.5×105/kg as required by the planned therapeutic study protocol. T cell content was reduced by 1.07 log (0.84–3.17 log) at the expense of a significant NK cell loss of 47% (32–76%). Bacterial contamination tests from all culture bags were negative. These results demonstrate the feasibility of clinical grade large-scale ex vivo NK cell expansion. GMP-expanded NK cells exhibited high levels of activating receptors NKG2D and NKp44 (3.1 and 16.8 fold increase above basal levels). The proportions of alloreactive NK cells with single killer immunoglobulin-like receptor (KIR) specificities remained stable comprising 2.4–16.0% and 5.2–14.2% of cells before and after expansion. GMP-expanded NK cells were more cytotoxic against K562 target cells than freshly isolated NK cells (mean±SEM: 61.0±3.1% vs. 27.0±8.0% specific lysis at 10:1 effector to target ratio). The cytolytic activity against KIR-HLA class I mismatched primary AML blasts was 7.0±3.7%, reflecting the content of alloreactive NK cell subsets. This is supported by our finding that purified single-KIR alloreactive NK cells lysed mismatched AML blasts with a high efficiency (22.0±6.3% at 10:1 ratio), indicating the anti-leukemic capacity of GMP-expanded NK cells. Two NK-DLI products with a total of 21.0×108 and 189.4×108 NK cells were generated for two patients and divided into three escalating doses (1×106, 1×107 and 1.5×107/kg) for patient 1 and into four doses (1×106, 1×107 and twice 1×108/kg) for patient 2. The infusions were tolerated without any acute adverse effects. In conclusion, we established and clinically implemented a protocol suitable for GMPcompliant expansion of cytokine-activated NK cells allowing multiple infusions of high numbers of cells with strong cytolytic activity. These results provide the basis for a prospective clinical efficacy trial to advance the therapeutic field of NK-DLI against human cancer.