Imaging of cell-based therapy using ⁸⁹Zr-oxine ex vivo cell labeling for positron emission tomography

Abstract

With the rapid development of anti-cancer cell-based therapies, such as adoptive T cell therapies using tumor-infiltrating T cells, T cell receptor transduced T cells, and chimeric antigen receptor T cells, there has been a growing interest in imaging technologies to non-invasively track transferred cells in vivo. Cell tracking using ex vivo cell labeling with positron emitting radioisotopes for positron emission tomography (PET) imaging has potential advantages over single-photon emitting radioisotopes. These advantages include intrinsically higher resolution, higher sensitivity, and higher signal-to-background ratios. Here, we review the current status of recently developed Zirconium-89 (⁸⁹Zr)-oxine ex vivo cell labeling with PET imaging focusing on its applications and future perspectives. Labeling of cells with ⁸⁹Zr-oxine is completed in a series of relatively simple steps, and its low radioactivity doses required for imaging does not interfere with the proliferation or function of the labeled immune cells. Preclinical studies have revealed that ⁸⁹Zr-oxine PET allows high-resolution in vivo tracking of labeled cells for 1-2 weeks after cell transfer both in mice and non-human primates. These results provide a strong rationale for the clinical translation of ⁸⁹Zr-oxine PET-based imaging of cell-based therapy.

Introduction

The last decades of development in cell-based anti-cancer therapies has been marked by breakthroughs such as adoptive transfer of tumor-infiltrating T cells (TIL) ^{1, 2}, tumor antigen specific T cell receptor (TCR)-transduced T cells (TCR-T) ^{3, 4}, and chimeric antigen receptor T (CAR-T) cells ^5-9. Adoptive transfer of TIL in melanoma patients achieves objective responses of 50 - 70 % and complete responses of 22 % when a lymphodepleting preparative regimen is employed before transfer ^{1, 2, 10, 11}. CD19-directed CAR-T cells achieve durable complete responses in 40 % of refractory diffuse large B cell lymphoma and in 70 % of follicular lymphoma. However, some severe, sometimes fatal, side effects, including cytokine release syndrome that correlates with tumor burden and encephalopathy, have been observed ^{9, 12}. Along with the evolution of CAR-T cells with improved intracellular signal transduction to enable proliferation and cytokine production of T cells ⁹, CAR technology has been expanded to other cell types such as B cells and NK cells ¹³. Other major breakthroughs have been made in the field of stem cell therapies. In addition to traditional hematopoietic stem cell transplants, mesenchymal stem cells or cells differentiated from induced pluripotent stem cell (iPSC) developed from the patient's own cells or embryonic stem cells have been actively investigated for the therapy or regeneration of damaged tissue ^{14, 15}.

Common to all the cell-based therapies is their dependency on the delivery of transferred cells to the desired target in order for the cells to exert therapeutic effects. Even in the case of CAR-T cells that have enhanced target antigen recognition capabilities, the cells first have to migrate to and then infiltrate the target (e.g. cancer) tissue in order for them to recognize the specific antigen and exert cytotoxicity ^16-18. The accumulation of transferred tumor-targeting lymphocytes can vary among the metastatic lesions within the same patient as a result of different evolutionary and/or immunoediting processes unique to each metastatic lesion ^{19, 20}. In most clinical cell-based therapies, biopsies or observations of clinical outcome have been the method to evaluate the migration of transferred cells to the target tissue. Even if tumors are biopsied, transferred lymphocytes often cannot be distinguished from endogenous tumor-infiltrating lymphocytes unless transferred cells are labeled in advance ^{21, 22}. The number of cells reaching the target are often quite small. Therefore, imaging of transferred cells is clinically important in order to non-invasively and quantitatively evaluate the homing of the cells to the targeted cancers but also to use as a tool in designing next generation cell therapies with better homing characteristics.

Technically, transferred cells can be tracked using single-photon emission computed tomography (SPECT) or positron emission tomography (PET) when the cells are labeled with appropriate radioisotopes. Indium-111 (¹¹¹In)-oxine or Technetium-99m (^99mTc)-hexamethylpropylene amine oxime (HMPAO) scintigraphy/SPECT has been a classical method to visualize leukocytes and is clinically used for evaluation of inflammation, infection or abscess since the mid-1970s ^23-26. ¹¹¹In-oxine scintigraphy/SPECT has been previously used to analyze the distribution of adoptively transferred TILs in patients ^{21, 22}. However, the sensitivity of SPECT is 2 to 3 orders of magnitude lower than that of PET ²⁷, and relatively high doses of radioactivity are required for imaging, which cause death or dysfunction in the labeled cells ^28-30. ^99mTc has a short half-life (6 hours) and is not suitable for tracing the transferred cells for multiple days, and cell labeling with ^99mTc-HMPAO is relatively unstable ^{30, 31}. Alternatives to radiolabeling include magnetic resonance imaging (MRI) using superparamagnetic iron oxide nanoparticles (SPION) that are phagocytized by the cells to be tracked ³². While MRI provides detailed anatomical information without ionizing radiation, it is cumbersome to survey the entire body and also difficult to detect the signal loss caused by iron oxide unless pre- and post-SPION loading studies are compared. This poses practical concerns as well. Moreover, quantitation of SPION's dark signal void is difficult. One of the recent developments in the tracking of cells with MRI is to use Fluorine-19 (¹⁹F) labeling. Here, the MRI unit is tuned to the resonance frequency of ¹⁹F, which blocks out all the background signal from water protons. However, this method requires specialized ¹⁹F detection coils, and because of limitations in sensitivity, large amounts of ¹⁹F have to be introduced to the cells or tissues to generate detectable signals ^{32, 33}.

In general, PET is superior in image quality, sensitivity, spatial resolution, and quantification to SPECT ²⁷, which makes it an attractive modality for tracking cells. Among positron (β)-emitting radioisotopes used for PET, ⁸⁹Zr has a relatively long half-life (3.3 days), ideal for tracking cells for several days, and relatively low positron energy that is required for high resolution in PET ^{34, 35}. Therefore, PET imaging using ⁸⁹Zr has been gaining attention, and the usefulness of ⁸⁹Zr-immunoPET, in which antibodies or antibody fragments are labeled with ⁸⁹Zr (e.g. ⁸⁹Zr-labeled anti-PD-1/PD-L1 antibody), for imaging human patients has been evaluated ^36-39. For the specific task of tracking adoptively transferred cells for multiple days with relatively high resolution, cells to be tracked are labeled with ⁸⁹Zr-oxine and imaged by PET ^{40, 41}. Here, we briefly review the current status of cell labeling and the applications of ⁸⁹Zr-oxine PET imaging technique for cell-based therapy.

Methods to track cells in vivo by PET imaging

Radionuclides that have been used for ex vivo cell labeling to longitudinally evaluate the distribution of the cells of interest in vivo include; Fluorine-18 (t_1/2 = 109.7 minutes), in the form of ¹⁸F-fluorodeoxyglucose (FDG) to label cells ex vivo before transfer ⁴², Copper-64 (⁶⁴Cu, t_1/2 = 12.7 hours), such as copper-64-pyruvaldehyde-bis (N4-methylthiosemicarbazone) (⁶⁴Cu-PTSM) ⁴³, and ⁸⁹Zr and Iodine-124 (t_1/2 = 4.2 days), in the form of radio-labeled antibodies against specific cell surface proteins ⁴⁴. ¹⁸F-FDG depends on glucose uptake by the cell and thus is not suitable for dormant cells ⁴⁵. ¹⁸F-FDG also suffers from relatively rapid efflux after labeling, due to dephosphorylation of ¹⁸F-FDG-6-phosphate, the phosphorylated form of ¹⁸F-FDG that will be trapped within the cell ^{42, 45, 46}. ⁶⁴Cu-SPION has been recently developed as a PET-MRI multi-modal imaging nanoparticle that has potential benefits for tracking cells in vivo, taking advantage of both the high sensitivity of PET and detailed anatomical information of MRI ⁴⁷. However, the half-life of ⁶⁴Cu, although longer than that of ¹⁸F, is still insufficient for tracking cells beyond several days after transfer. On the other hand, ⁸⁹Zr, with a half-life of 3.3 days, is ideal for cell tracking for 1-2 weeks. Its relatively lower positron energy improves PET spatial resolution and thus, is most often used for studies of this type.

In in vivo cell labeling, a radioactive labeling reagent (e.g. radioisotope-conjugated antibodies, antibody fragments, or engineered antibodies for immunoPET) is systemically administered and the distribution of the target molecule expressed on the endogenous cell populations in the body is evaluated ^36-39. Thus, in vivo labeling cannot discriminate adoptively transferred cells from the endogenous cells expressing the same target molecules (e.g. adoptively transferred CD8 T cell versus endogenous CD8 T cells). Therefore, although in vivo cell labeling is useful for analyzing the distribution of endogenous cell populations ^{38, 39}, it is not suitable for the evaluation of the fraction of adoptively transferred cells that reach the target tissue. In vivo labeling is frequently associated with high background signal due to specific or non-specific antibody distribution (e.g. liver and spleen) and clearance (e.g. renal clearance of antibody fragments), long circulation time in the blood and pooling in tissue fluid, including those caused by enhanced permeability and retention (EPR) effect in some tumor tissues ⁴⁸. In vivo labeling also requires highly specific antibodies with little to no cross-binding to non-target cells; however, specific cell markers, although present on many T and B cells, are frequently unavailable for many other cell types such as dendritic cells or monocytes.

If adoptively transferred cells are transduced with a PET reporter gene, it is possible to administer a radioactive reporter probe to locate the transferred cells in the body ⁴⁹. Since the expression level of the reporter gene is maintained after cell proliferation, reporter gene imaging can achieve longer-term tracking of transferred cells. It also detects only live cells. On the other hand, the requirement for an injection of the radioactive probe and clearance of unbound probe at each interested time point limits short-term or frequent monitoring of cell migration. Thus, monitoring cellular responses to additional stimuli that cause relatively rapid changes of cell distribution could be difficult. The biggest disadvantage of PET reporter gene imaging is its requirement for gene introduction to the cells. It is imperative that such reporter genes are carefully chosen to avoid immunogenicity. Although PET reporter gene imaging is feasible for therapeutic cells that undergo genetic engineering during their production, such as CAR-T cells, for cells that do not require genetic engineering, which are the majority of TILs and stem cells, reporter gene imaging strategies would be difficult to translate into the clinic.

In contrast to in vivo labeling, cells can be labeled with radioisotopes ex vivo before adoptive transfer to patients ^{40, 41}. Since only the cells adoptively transferred are labeled with radioisotope, ex vivo labeling has essentially zero background signal in the recipient, thereby achieving extremely high signal-to-background ratios even with low cell labeling doses, and, therefore, high sensitivity to even tiny clusters of labeled cells ^{40, 41}. Monitoring relatively small cell distribution changes to additional stimuli could be possible with this method. Ex vivo labeling can accurately track adoptively transferred cells. However, some background signals may develop over time due to phagocytosis of damaged labeled cells or release of the radioactive label after death of the labeled cells (e.g. accumulation of ⁸⁹Zr within bone as described below). As each cell division results in approximately halving of the label per cell, image intensity can decrease due to cell division alone ⁴⁰.

Labeling of cells with ⁸⁹Zr-oxine for PET imaging

Cell labeling by ⁸⁹Zr-oxine is completed in three relatively simple steps that can be performed in aqueous solution ⁴⁰: 1) synthesis of ⁸⁹Zr-oxine by mixing ⁸⁹Zr-chloride with oxine followed by neutralization at room temperature. ⁸⁹Zr-chloride can be generated from ⁸⁹Zr-oxalate produced by a cyclotron ⁵⁰, 2) incubation of cells with synthesized ⁸⁹Zr-oxine for 15 minutes at room temperature or at 4 ^oC, and 3) washing of the labeled cells. The ⁸⁹Zr-oxine synthesis yield is as high as >98 %, which, together with the aqueous conditions, eliminates the need for a purification step. With application of the generated ⁸⁹Zr-oxine solution to a cell suspension, virtually all the cells are labeled with ⁸⁹Zr-oxine. The synthesis and labeling steps are similar to those of ¹¹¹In-oxine complex and thus, ⁸⁹Zr-oxine cell labeling could be easily translated to the clinic. Recently, a slightly modified method of generation of ⁸⁹Zr-oxine, which uses ⁸⁹Zr-oxalate instead of ⁸⁹Zr-chloride, has been reported ⁵¹. Another reported method of synthesis uses neutralized ⁸⁹Zr-oxalate to react with oxine in chloroform, followed by recovery of the synthesized ⁸⁹Zr-oxine from chloroform phase by evaporation and re-dissolving the obtained ⁸⁹Zr-oxine in dimethyl sulfoxide for further dilution in aqueous buffer for cell labeling ^{40, 41}.

Mechanistically, the lipid solubility of ⁸⁹Zr-oxine allows it to diffuse through the cell membrane. Therefore, labeling with ⁸⁹Zr-oxine does not require active uptake by cells and can be performed in resting cells and at 4 ^oC ⁴⁰. This characteristic also eliminates the requirement for expression of certain cell surface molecules for labeling or an antibody specific cell surface protein.

Experience with ¹¹¹In-oxine is instructive. 8-hydroxyquinoline (oxine) can be released from ¹¹¹In-oxine after the molecule enters the cytoplasm, and the free ¹¹¹In forms a stable complex with cytoplasmic and nuclear proteins ⁵². The formation of ¹¹¹In-protein complex is presumably achieved through exchange of ¹¹¹In from 8-hydroxyquinoline to intracellular proteins ⁵². Free 8-hydroxyquinoline is then exported from the cell. Although a similar mechanism can be assumed for ⁸⁹Zr-oxine, the exact mechanism of cell labeling using ⁸⁹Zr-oxine is still unclear. Our preliminary results have shown that ⁸⁹Zr-oxine also labels the cell membranes (cytoplasmic and intracellular membranes) as well as cytoplasmic, and nuclear proteins, while negligible binding was observed with chromatin and cytoskeleton proteins (unpublished data). Although ⁸⁹Zr-oxine can label a variety of cells, the labeling efficiency (percentage of ⁸⁹Zr incorporated to the cells per added activity) differs among cell types ⁴⁰. This may be due to the difference in the amount of intra-cellular proteins available for ⁸⁹Zr binding. ⁸⁹Zr-oxine labeled cells, especially non-dividing cells (e.g. matured dendritic cells), retain ⁸⁹Zr very well over multiple days ⁴⁰.

In contrast to the cell-permeable ⁸⁹Zr-oxine, ⁸⁹Zr-deferoxamine-NCS has been reported to directly conjugate ⁸⁹Zr to cell surface proteins for cell tracking by PET ^{53, 54}. In this method, deferoxamine chelates ⁸⁹Zr, while the NCS group covalently binds primary amine groups of membrane proteins on the cell surface. Therefore, ⁸⁹Zr radioactivity is mainly detected in the membrane fraction of labeled cells but not in cytoplasmic and nuclear fractions ⁵³.

Maintenance of immune cell function after labeling with ⁸⁹Zr-oxine

Importantly, labeling of immune cells including lymphocytes, macrophages, and bone marrow-derived dendritic cells with ⁸⁹Zr-oxine does not affect their function when optimal labeling doses were used. For example, labeling with ⁸⁹Zr-oxine does not interfere with the activation, proliferation, and cytokine expression (IFN-γ and IL-2) after TCR stimulation in labeled CD8 T cells ⁴⁰. Similarly, labeling of CAR-T cells by ⁸⁹Zr-oxine does not affect their viability, cytotoxicity, or cytokine production ⁵¹. Labeling with ⁸⁹Zr-oxine does not interfere with LPS-induced activation of dendritic cells and their ability to present antigens and activate T cells ⁴⁰. The only exception observed to date, is that ⁸⁹Zr-oxine labeling of bone marrow cells, even at low radioactivity doses (10.9 kBq/10⁶ cells), showed initial delay in the proliferation after granulocyte colony stimulation factor (GM-CSF) stimulation in vitro ⁵⁵. However, the differentiation capacity of these labeled bone marrow cells is maintained both in vitro and in vivo ⁵⁵. Based on these studies, the optimal labeling doses minimize cellular toxicity while being sufficient to image by PET are approximately 11 - 44 kBq/10⁶ cells, depending on type and condition of the cells.

Once labeled with ⁸⁹Zr-oxine, ⁸⁹Zr is stably retained in labeled cells ⁴⁰. However, death of labeled cells results in rapid release of free ⁸⁹Zr. In vivo, free ⁸⁹Zr is excreted from the kidney; however, a small fraction of free ⁸⁹Zr may be taken up in bone matrix hydroxyapatite ^{40, 55, 56}. This small uptake in bone can be reduced by administration of deferoxamine, a chelating agent clinically used for treatment of iron overload and hemochromatosis ⁵⁵. Once bound to deferoxamine, the ⁸⁹Zr is rapidly excreted in the urinary tract.

Monitoring of transferred cells using ⁸⁹Zr-oxine PET in mouse models

Preclinical studies have shown that ⁸⁹Zr-oxine PET can track adoptively transferred cells over a period of 1-2 weeks. For example, adoptive transfer of ⁸⁹Zr-oxine-labeled activated OT-I CD8 T cells (TCR transgenic T cells that recognize ovalbumin: OVA) shows initial rapid accumulation in the lungs, followed by gradual accumulation within antigen-expressing tumor (B16-OVA) by Day 2 (Figure 1A, right) ⁴⁰. In contrast, these activated OT-I CD8 T cells do not accumulate in antigen-negative B16 tumors. Uptake is restricted to B16-OVA tumors, which is associated with a rapid anti-tumor effect (Figure 1B). In addition to the antigen-expressing B16-OVA tumor, ⁸⁹Zr-oxine PET shows that activated OT-I CD8 T cells also distribute to the liver after adoptive transfer, compatible with previous observations that activated CD8 T cells preferentially home to the liver ^{57, 58}. On the other hand, ⁸⁹Zr-oxine PET demonstrates that naïve CD8 T cells preferentially distribute to the spleen and lymph nodes instead of the liver (Figure 1A, left). Therefore, ⁸⁹Zr-oxine PET clearly captures the different behaviors between naïve and activated CD8 T cells after adoptive transfer. Similar results have been observed when ⁸⁹Zr-oxine-labeled CAR-T cells are used, wherein transferred prostate stem cell antigen (PSCA)-targeting CAR-T cells accumulate in the liver as well as the PSCA-expressing tumor, after initial distribution to the lungs ⁵¹.

Figure 1.

Tracking of transferred cells using ⁸⁹Zr-oxine PET in murine models. A) Activated OT-1 CD8 T cells distribute to antigen-expressing B16-OVA tumor (white dashed circle), while naïve OT-1 CD8 T cells do not. Naïve OT-1 CD8 T cells (left) and activated OT-1 CD8 T cells (right, adapted from ⁴⁰) were labeled with ⁸⁹Zr-oxine (222 kBq/8 x 10⁶ cells and 248.5 kBq/7.7 x 10⁶ cells, respectively) and intravenously transferred to tumor-bearing recipient mice. PET/CT images acquired on 2nd day after transfer are shown. Also note the different distribution between naïve and activated OT-1 CD8 T cells outside the tumor. B) Activated OT-1 CD8 T cells distribute to antigen-expressing B16-OVA tumor (cyan circle) but not to antigen-negative B16 tumor (white dashed circle). Activated OT-1 CD8 T cells labeled with ⁸⁹Zr-oxine (185 kBq/8 x 10⁶ cells) were intravenously transferred to the recipient. MicroPET/CT images acquired on 4th day after transfer are shown. Left: maximum intensity projection MIP PET/CT, right: transverse plane. C) Distribution of transferred bone marrow cells. Donor bone marrow cells were labeled with ⁸⁹Zr-oxine (16.6 kBq/2.0 x 10⁷ cells) and intravenously transferred to the recipient. PET/CT imaging was performed at indicated time points after transfer, with multiple intramuscular injections of deferoxamine to prevent accumulation of free ⁸⁹Zr within bone. Transferred bone marrow cells initially distribute to the lungs, followed by distribution within 4 hours to the bone marrow, liver, and spleen. Adapted from ⁵⁵.

Vγ2Vδ9 subtype of γδ T cells, which represent most circulating γδ T cells in humans, can recognize phosphoantigens expressed on tumor cells and exert anti-tumor activity ⁵⁹. When ex vivo expanded human Vγ2Vδ9 T cells are labeled with ⁸⁹Zr-oxine and transferred to tumor-bearing SCID mice, ⁸⁹Zr-oxine PET shows accumulation in the tumor and liver, after transient distribution to the lungs. ⁶⁰. This can be detected by ⁸⁹Zr-oxine PET and is augmented by administration of aminobisphosphonate, which increases the expression of phosphoantigens on the surface of tumors ^{59, 61}.

Tracking of bone marrow cells also reveals interesting patterns. ⁸⁹Zr-oxine PET shows the initial transient distribution of transferred bone marrow cells to the lungs, rapidly followed by homing to the bone marrow, spleen, and liver within 4 hours of transfer (Figure 1C) ⁵⁵. Since the homing of bone marrow cells is dependent on CXCL12-CXCR4 axis ⁶², ⁸⁹Zr-oxine PET shows delayed homing of labeled donor bone marrow cells to recipient bone marrow after administration of plerixafor, a CXCR4 antagonist ⁵⁵. Co-administration of plerixafor and G-CSF further disturbed this homing process ⁵⁵. On the other hand, ⁸⁹Zr-oxine PET shows that the initial homing of transferred bone marrow cells is not affected by prior whole body irradiation ⁵⁵. In a bone fracture model, ⁸⁹Zr-oxine PET further shows the mobilization of pre-transferred ⁸⁹Zr-labeled bone marrow cells from pre-distributed organs (bone marrow, spleen, and liver) to a site of fracture within 1 day of injury ⁶³. This re-distribution of pre-distributed bone marrow cells is also impaired by plerixafor treatment ⁶³. These reports indicate that ⁸⁹Zr-oxine labeling of highly radiosensitive bone marrow could be performed with minimum effect on functionality of the cells.

With the ⁸⁹Zr-deferoxamine-NCS direct membrane protein conjugation method, human mesenchymal stem cells have been imaged, showing their distribution to the lungs and liver after intravenous injection in athymic nude mice ⁵³. Interestingly most human mesenchymal stem cells labeled with ⁸⁹Zr-deferoxamine-NCS remain within the injected site for at least one week after injection into myocardium in athymic nude mice ⁵³.

Monitoring of transferred cells using ⁸⁹Zr-oxine PET in non-human primates

⁸⁹Zr-oxine PET can be used to image adoptively transferred cells in non-human primates. For example, in an autologous intravenous hematopoietic stem cell transplantation model in rhesus macaques, ⁸⁹Zr-oxine PET imaging shows that CD34⁺ hematopoietic stem and progenitor cells (HSPCs) accumulate in the bone marrow, liver, and spleen after rapid passage through the lungs (Figure 2A) ⁶⁴, similar to the pattern seen in a murine model of bone marrow transplantation (Figure 1C) ⁵⁵. Clinically, intrabone transplantation of HSPCs, especially cord blood cells, has been suggested as a method to improve engraftment in the recipient's bone marrow ^{65, 66}, and preclinical studies using swine and macaque were undertaken to determine if homing and engraftment of HSPCs was improved with intrabone transplantation compared to intravenous transfer ^{67, 68}. ⁸⁹Zr-oxine PET was used to further confirm the retention of HSPCs at the injection site in the iliac bone in rhesus macaques ⁶⁹.

Figure 2.

Tracking of transferred cells using ⁸⁹Zr-oxine PET in non-human primate models. A) Distribution of transferred CD34⁺ hematopoietic stem and progenitor cells (HSPCs) in a rhesus macaque. CD34⁺ HSPCs mobilized by plerixafor and G-CSF treatment were labeled with ⁸⁹Zr-oxine and autologously transferred intravenously (44.4 kBq/10⁶ cells, 1.2 x 10⁶ cells/kg). PET/CT was performed longitudinally under continuous infusion of deferoxamine to prevent accumulation of free ⁸⁹Zr within bone. Transferred cells rapidly distribute to the bone marrow, liver, and spleen ⁶⁴. B) Distribution of transferred NK cells in a rhesus macaque. NK cells purified from peripheral blood were expanded ex vivo with IL-2, labeled with ⁸⁹Zr-oxine, and autologously transferred intravenously (15.2 kBq/10⁶ cells, 21.9 x 10⁶ cells/kg). Cell migration was longitudinally tracked by PET/CT. Deferoxamine was continuously infused during the entire imaging study. Transferred NK cells initially distribute to the lungs, followed by gradual distribution to the liver and spleen. Little distribution of NK cells to the bone marrow is observed. Adapted from ⁶⁴.

In another study, NK cells were expanded ex vivo with IL-2, per protocol used for human NK cell clinical trials, labeled with ⁸⁹Zr-oxine, and tracked by ⁸⁹Zr-oxine PET after autologous adoptive transfer, for preclinical evaluation of NK cell therapy against hematological malignancies such as leukemia. In this study, ⁸⁹Zr-oxine PET revealed that transferred NK cells showed only minimal homing to the bone marrow, and most transferred NK cells distribute to the liver and spleen (Figure 2B) ⁶⁴. This implies that for NK cells to be useful in treating bone marrow malignancies, NK cells must be engineered to track to the bone marrow in greater numbers than unmodified NK cells.

In humans, it has been known since the 1940s that glucocorticoid treatment rapidly decreases the number of eosinophils in peripheral blood within 1-2 hours of administration ⁷⁰; however, it has been unclear where the eosinophils re-distribute in the body. To address this question, eosinophils purified from peripheral blood were labeled with ⁸⁹Zr-oxine, transferred back to the animal, and imaging was performed before and after glucocorticoid treatment. ⁸⁹Zr-oxine PET demonstrated a rapid increase of bone marrow accumulation of ⁸⁹Zr-oxine labeled eosinophils over a four-hour period after glucocorticoid treatment ⁷¹. This demonstrates the high sensitivity of ⁸⁹Zr-oxine PET to detect rapid changes in cell distribution in response to exogenous stimuli.

These results indicate that high-resolution cell tracking using ⁸⁹Zr-oxine PET can also be successfully performed in non-human primates using clinical PET scanners, providing a rationale for the translation of cell tracking using ⁸⁹Zr-oxine PET to human trials.

Pitfalls and clinical translation of ⁸⁹Zr-oxine PET

There are several pitfalls associated with ⁸⁹Zr-oxine PET and other ex vivo labeling methods. One is the possible transfer of the label to phagocytes after the death of labeled cells. As observed in the use of iron nanoparticles for MRI ⁷², radioisotopes that are tightly bound to a cell structure or deposited in tissue could be transferred to tissue monocytes/macrophages by phagocytosis, resulting in non-specific signals. In this regard, studies using ⁸⁹Zr-oxine so far have shown that ⁸⁹Zr is released from dead cells and mostly cleared from the kidney. This process can be expedited by utilizing deferoxamine infusion as described below ^{55, 64}. ⁸⁹Zr-oxine does not passively transfer from labeled cells to neighboring cells ⁶⁴.

Another pitfall is that there can be bone uptake from small amounts of released ⁸⁹Zr after death of labeled cells ^{40, 55, 56}. However, this can be again avoided by deferoxamine to chelate free ⁸⁹Zr and enhance clearance from kidneys ^{55, 64}. Deferoxamine is clinically used for iron overdose and thus potentially could be used safely during the imaging.

Before translation of ⁸⁹Zr-oxine PET technology into the clinical setting, a number of important criteria must be met. There must be simplicity in the labeling method, low cytotoxicity in labeled cells, and no side effects to recipients of labeled cells. Cell labeling with ⁸⁹Zr-oxine is straightforward and exerts minimal toxicity to labeled cells when optimal cell labeling doses were used. Although no clinical trials in humans have been performed, dosimetry data in non-human primates that received autologous NK cells showed very low radio-exposures to organs ⁶⁴, and no clinical side effects were observed with autologous transfer of NK cells, CD34⁺ HPSCs ⁶⁴ and eosinophils ⁷¹, confirming the safety of this method. Practically, labeling only a small fraction (e.g. 10 %) of the cells infused for cell-based therapy should be sufficient to obtain PET images with sufficiently high quality for cell tracking.

Concluding remark

Here, we have reviewed our experience with cell labeling using ⁸⁹Zr-oxine in preclinical models of in vivo tracking. Cell labeling with ⁸⁹Zr-oxine can be performed in relatively simple steps. Once labeled, ⁸⁹Zr is stably retained in cells until cell death. After cell death, ⁸⁹Zr is quickly liberated and generally cleared from the kidneys with enhanced clearance if deferoxamine is infused. ⁸⁹Zr-oxine PET allows high-resolution tracking of transferred cells both in murine and non-human primate models for 1-2 weeks after transfer. These preclinical observations collectively provide a strong rationale for clinical translation of this technology.

Abbreviations

CAR-T

chimeric antigen receptor T

HSPC

hematopoietic stem and progenitor cell

iPSC

induced pluripotent stem cell

MRI

magnetic resonance imaging

NK

natural killer

OVA

ovalbumin

PET

positron emission tomography

SPECT

single photon emission computed tomography

SPIO

superparamagnetic iron oxide

TCR

T cell receptor

TIL

tumor-infiltrating lymphocyte

⁸⁹Zr

Zirconium-89