Background: Clinical adoption of NK cell immunotherapy is underway for pediatric medulloblastoma and osteosarcoma [1,2]. The first Phase I clinical trial of autologous NK cell therapy in pediatric patients with recurrent, refractory, metastatic fourth ventricular tumors showed no adverse events, with over 112 infusions of NK cells and doses up to ~3 x 10⁸ NK cells [1]. This study underscored the need for tools to assess NK cell delivery and localization within the tumor microenvironment. Magnetic particle imaging (MPI) is an emerging modality for in vivo tracking of cell therapies [3-6] and clinical-scale scanners are under development worldwide [7-9].

Objectives: Our first objective is to label NK cells with iron oxide nanoparticles to enable MPI detection without impacting cell viability or activity. We aim to detect, quantify, and measure the location of labeled NK cells following surgical implantation to mouse cerebellum and tibias using preclinical MPI. Our final objective is to demonstrate the feasibility for detection of a clinically relevant number of labeled NK cells on a human-scale MPI scanner.

Materials and Methods: Human derived NK cells (NK-92, ATCC) were labeled by co-incubation with 50 or 200 µg Fe/mL Iron Oxide (VivoTrax, Magnetic Insight) for 24 hours then the excess nanoparticles were washed with media by centrifugation. Cell labeling was assessed qualitatively with Perl’s Prussian blue (PPB) staining and quantitatively using a colorimetric iron assay kit (MAK025, Sigma). Cytolytic activity of labeled versus unlabeled cells was assessed after 4 hours of co-incubation with DAOY medulloblastoma cells or LM7 osteosarcoma cells using bioluminescent or GFP counts. Unlabeled or iron-oxide-labeled NK cells at two different doses (0.5 or 1 x 10⁶) were injected by stereotactic administration to excised mouse brains (cerebellum) and legs (tibias), to match the sites of medulloblastoma and osteosarcoma. Mouse brains and legs with injected cells were imaged by preclinical MPI (Momentum, Magnetic Insight) in 2D (2-minute acquisition) and 3D (30 minutes). A clinically relevant dose (40 x 10⁶) of iron-oxide-labeled cells was imaged using a prototype clinical-scale MPI with a coil of inner diameter = 25 cm to produce an image with dimensions 20 x 20 x 3 cm (5.5 minutes).

Results: PPB histology verified intracellular cell labeling. Following incubation with 200 µg Fe/mL, cells were labeled with an average of 3.17 pgFe/cell with no measurable effects on cell viability or cytolytic activity against DAOY and LM7 cancer cell lines. The preclinical MPI detection limit was determined as 3.1 x 10⁴ NK cells (cell pellet). MPI signal was directly quantitative with cell number for cell pellets (R² = 0.97) and following implantation to mouse brains and tibias. No background MPI signal is detected from mouse brains or legs that received injections of unlabelled cells, underscoring the high specificity of MPI signal. Using 3D MPI and fiducials as anatomical markers, the administration site of labeled cells could be accurately measured in mouse brains within 0.1-0.6 mm of target stereotactic injection coordinates. Further, MPI resolution was sufficient to visualize that cells had been administered to two focal regions in the tibia, separated by 4.2 mm. We demonstrate detection of 40 million iron-oxide-labeled cells using human MPI scanner with peak SNR = 110.

Conclusions: MPI can provide specific, quantitative, and accurate spatial information on iron-labeled NK cells soon after delivery, showing initial promise to address a significant unmet clinical need to track cell fate in patients. MPI shows high sensitivity to detect iron-labelled cells on a human-scale MPI for the first time. Future work will focus on the potential of MPI to track in vivo longitudinal NK cell efficacy in tumor-bearing mice.

Image/Figure:

Click to view full size

Image/Figure Caption:

(a,b) Perl’s Prussian blue histology confirms NK cells were successfully labeled with VivoTrax following co-incubation with 50 µg Fe/mL (11% labeled cells) and 200 µg Fe/mL (26% labeled cells). NK cell function was retained following VivoTrax labeling as shown by cytolysis assays at various effector:target ratios of (c) DAOY-ffluc medulloblastoma cells and (d) LM7-GFP osteosarcoma cells. (e) VivoTrax-labeled NK cells (4 x 10⁷ cells) were detected on a human-scale MPI scanner producing a 3D image with dimensions 20 x 20 cm x 3 cm. (f) 3D MPI of ex vivo mouse brain showing localization of 1 x 10⁶ VivoTrax-labeled NK-92 cells implanted to right cerebellum relative to 3 fiducials (F1, F2, F- λ). (g) 2D MPI of ex vivo mouse legs showing increasing amount of MPI signal from unlabeled, 0.5×10⁶ labeled and 1.0×10⁶ labeled NK-92 cells implanted to the tibia (* p < .05, ** p < .01, **** p < 0.0001). (h) A dilution series of VivoTrax-labeled NK cells, T cells, and macrophages labeled using the same protocol imaged by preclinical MPI. The limit of detection was determined as 31×10³ (31k) for NK cells and T cells and 8×10³ (8k) for macrophages. For each cell type, cell number was directly linear with MPI signal produced (R² > 0.96).

Author