Here we present first interesting insights from our clinical NK cell phase I/II study using allogeneic NK-DLIunstim
compared to NK-DLIIL-2 stim
in pediatric patients suffering from high risk malignancies. Although, we, among others, have shown that the infusion of unstimulated as well as previously ex vivo
IL-2 stimulated allogeneic NK cells post haplo-SCT is well tolerated without inducing severe GvHD>grade II 
, possible risks or disadvantages need to be critically discussed. As it stands, literature is scarce about the fate and behavior of adoptively transferred allogeneic NK cells in humans and about the potential distinct influences of unstimulated NK cells in contrast to previously ex vivo
activated NK cells on patient's adaptive and innate immune system.
Several studies in animals have addressed the question concerning the capability to traffic to specific tissues, the regulation of homing, and the survival of adoptively transferred cells in vivo
. NK cell trafficking to spleen, lymph nodes, lung, liver, gastrointestinal tissue and tumor side with a survival up to four weeks following transfer was observed by a bioluminescence-based strategy, which correlated with an observed anti-tumor effect 
. However, to date only one small clinical trial in humans was performed, where three adult patients with renal cell carcinoma received stimulated allogeneic NK cells labeled with the radioactive substance Indium-111 oxine 
. After an initial accumulation in the lungs, NK cells redistributed to liver, spleen and bone marrow as well as in two of four metastases in lung and liver. Unfortunately, it has been reported as well that Indium labeling significantly affects the cellular integrity 
. Even though it is still of particular interest if adoptively transferred NK cells in humans actually reach their side of action, clinical trials using NK cell labeling with potentially harmful substances will not obtain approval in the treatment of pediatric malignancies. Therefore, approaches using more noninvasive strategies have to be considered. Our investigation is based on a comprehensive in vivo
cytokine/chemokine monitoring and on flow cytometric analyses of quantification, constitution and distribution of various PB leukocyte subsets before and after NK-DLI application.
In our study we have reported markedly diverse effects between NK-DLIunstim and NK-DLIIL-2 stim. Shortly after infusion of NK-DLIIL-2 stim only, a rapid almost complete loss of cells dominantly from the innate immune system from patient's PB circulation appeared which was accompanied by significant increases in plasma concentration of various cytokines and chemokines. Whereas neutrophil granulocytes markedly increased within 4 h post NK-DLIIL-2 stim, monocytes, dendritic cells, eosinophils and especially NK cells massively decreased as early as 10 min post infusion, while recovering within the next 24 h.
Moreover, when analyzing NK cells more into detail, we were able to clearly discriminate between adoptively transferred and patients' PB NK cells by a distinct CD69, NCR and CD62L expression. We have shown previously that ex vivo
IL-2 stimulation leads to a predominantly CD56bright
phenotype with a strongly enhanced expression of the activation marker CD69, while CD62L becomes down-regulated. Further, surface receptors involved in NK cell cytotoxicity become highly up-regulated. While only one-third of unstimulated NK cells, a median of 95% of IL-2 stimulated NK cells show expression of NCRs. In detail, NKp44, NKp30, NKp46 and NKG2D expression significantly increased 33-fold, 12-fold, 3-fold and 4-fold, respectively 
. Furthermore, the IL-2 stimulation led to a consistent increase in NK cell killing activity against a neuroblastoma cell line 
and the leukemic cell line K562 (Fig. S2A
). In addition, Penack et al.
showed that the CD16−
NK cell subset is responsible for anti-tumor responses 
NK cells with the characteristics of the ex vivo IL-2 stimulated phenotype (CD56brightCD16+/−CD69+NCRhighCD62L−) were not detected in patients' PB at any time point during in vivo monitoring. Furthermore, we could clearly show that the significant reduction of CD56+CD3− NK cells from blood circulation following NK-DLIIL-2 stim was due to both, a decrease in patients' own PB CD62L+ NK cells as well as a rapid diminishing of the transferred, stimulated NK cells from the NK-DLI with the CD62L− phenotype.
In contrast, PB cell subpopulations remained constant after NK-DLIunstim. This effect was not due to NK-DLI dose, a PB dilution effect after infusion, application date or host's NK cell immune reconstitution. All these variables were very similar in both, patients receiving NK-DLIunstim and those receiving NK-DLIIL-2 stim. The only difference was the IL-2 for generation of NK-DLIIL-2 stim and the high amount of cytokines and chemokines such as IFN-γ, IL-8, MCP-1, IP-10, RANTES, MIP-1β secreted in the course of ex vivo expansion. Those factors were only transfused to patients treated with NK-DLIIL-2 stim.
In accordance with our results, early studies have reported a rapid diminishing of various types of PB lymphoid cells, especially NK cells, 15 min after in vivo
bolus single cytokine administration of very high doses of recombinant IL-2 (up to 1×106
U/kg BW). Similar to our study, cells also recovered within the next 24 h. Furthermore, IL-2 was rapidly cleared from the plasma with a half-life of 6.9 min 
. It has been suggested that the IL-2 induced disappearance of NK cells may be related to a massive adhesion to the activated endothelium 
. Our observed effects cannot be attributed to one single cytokine/chemokine but to the whole cytokine “cocktail” applied with the NK-DLIIL-2 stim
product, but the IL-2 dose applied by our NK-DLI study was extremely lower (~2×104
U/kg BW) in comparison to the discussed data by Lotze et al.
U/kg BW). Apparently, much lower concentrations of IL-2 but in combination with our indicated cytokines/chemokines administered by NK-DLIIL-2 stim
led to a comparable effect to high dose single IL-2 application with regard to PB leukocyte diminishing.
Measuring cytokine/chemokine production is an integral part of measuring immune response during immunotherapy. Because cytokines act in networks and have overlapping functions, monitoring of a single cytokine may be of limited use 
. Following NK-DLIIL-2 stim
we have shown significant increases in plasma concentration of several chemotactic and inflammatory cytokines and chemokines which remained enhanced up to 4 h post DLI (). The majority of the analyzed increases were probably induced by the infusion of high amounts of ex vivo
generated cytokines/chemokines in the NK-DLIIL-2 stim
. We assume that these changes in the natural cytokine milieu of the PB led to the observed cell migration processes. The massive increase of blood neutrophils, which represent the major early cell type to invade inflammatory foci, is likely mediated by the transfer of high amounts of IL-8 that were produced in the course of ex vivo
NK cell stimulation. Notably, IL-8 has been described to be the major chemo-attractant for neutrophil granulocytes. Neutrophils are described to be potent producers of various cytokines (i.e. IL-6, IL-8, IP-10, MIP-1α/β) which may be in relation to the prolonged enhanced cytokine/chemokine levels 4 h post NK-DLIIL-2 stim
Furthermore, the disappearance of various leukocyte subsets occurring only after NK-DLIIL-2 stim may be mediated by two alternative or complementary mechanisms: (i) adherence to the activated endothelium induced by high amounts of co-infused cytokines/chemokines, (ii) leukocyte migration from the PB into the extravascular compartment.
Although normal endothelial cells exhibit low affinity for circulating lymphocytes, the high amount of the cytokines and chemokines present in the PB (i.e. IFN-γ, MIP-1β, IL-8), similar to those released in the course of inflammation and other immune reactions, leads to endothelial activation associated with an increased expression of surface antigens which interact with all leukocytes 
. This might result in endothelial adherence and therefore diminishing of leukocytes from blood circulation. Further, it is known that soluble cytokines and chemokines bind endothelial molecules including glycosaminoglycans (GAGs) and the Duffy antigen/receptor for chemokines (DARC), which are involved in the trans-endothelial transport of several chemokines, i.e. MIP-1β, IL-8, RANTES, MCP-1 and IP-10 
. Chemokines bound at the luminal endothelial cell surface could provide a trans-cellular chemotactic gradient guiding leukocyte extravasation 
. Therefore, we assume that following the firm attachment to the activated endothelium, the cells migrate across the endothelium barrier into the tissue, actually leaving PB blood circulation. Although 7-AAD analyses revealed no increase in dead cells over the whole period of in vivo
monitoring (data not shown), a cell reduction in PB circulation due to cell death cannot be excluded completely.
In addition to the discussed trans-endothelial transport of cytokines/chemokines, cytokine stability in circulation, renal clearance as well as dilution effects during infusion must be regarded as parameters that have likely contributed to the described discrepancy between high ex vivo
levels in the NK-DLI and much lower in vivo
PB levels. In our study we found a significant reduction of the immune regulatory CD56bright
NK cell subpopulations 10 min after NK-DLIIL-2 stim
. An explanation for the overall higher susceptibility of the CD56bright
NK cell subpopulation might be the high expression of various chemokine receptors i.e. the MIP-1β corresponding CCR5 receptor on the cell surface of the CD56bright
subpopulation, only 
Finally, IL-6 was the only cytokine which was secreted much higher in the PB of our patients compared to those during NK cell expansion, leading us to suspect a secondary production of the patient's body in response to NK-DLIIL-2 stim. We speculate that this could be due to both, the secretion of endothelial cells in reaction to the changing cytokine milieu, neutrophil granulocytes and monocytes that transiently adhere to the endothelial surface. In addition, the increase of IL-6 in patients' blood plasma as a response to NK-DLIIL-2 stim correlated with our clinical observations of transient fever and chills; therefore serving as a surrogate marker of the biological activity of the ex vivo secreted and co-infused cytokines and chemokines.
Till now, very little is known about the effects of NK cell administration post SCT. These concomitant results to a clinical immunotherapy study provide first insights on the distinct influence of unstimulated vs. ex vivo IL-2 stimulated NK cell infusions. Nevertheless, we are fully aware that dissimilarities in the study design and the heterogeneous small patient cohort may have a potential effect on the results and that further studies have to verify the discussed data.
Moreover, an open issue remains the clinical benefit of NK-DLIIL-2 stim
compared to NK-DLIunstim
applications. Due to our heterogeneous patient cohort regarding different high risk diseases, with multiple and advanced relapses, mostly not in remission (NR), a clear evidence cannot be made. Anyhow, in the present study we could show a superior cytotoxicity of ex vivo
IL-2 stimulated compared to unstimulated NK cells against the MHC-I negative cell line K562 and against a neuroblastoma (NB) cell line as well 
. Cautiously it has to be noted, that 78% of the high risk group II patients treated with NK-DLIIL-2 stim
has not been in remission during haplo-SCT, but reached a survival of 44%. In addition in this group, two out of four patients suffering from high risk NB stadium IV with a very poor prognosis are still alive >2 years post NK-DLIIL-2 stim
, which seems promising and is in accordance with the enhanced lytic activity of IL-2 stimulated NK cells compared to NK-DLIunstim
against NB 
Conclusively, we were able to show that the adoptive transfer of NK-DLIIL-2 stim results in massive cell migrating processes under the influence of various ex vivo and most likely also in vivo secreted cytokines and chemokines. Since IL-2 activation leads to an improved cytotoxic capacity of the adoptively transferred NK cells, the co-transfused cytokine milieu may promote NK cell trafficking as well as an enhanced efficacy of NK cell immunotherapy.