Introduction
Based on insights from allogeneic stem cell transplantation and the practice of donor lymphocyte infusions, clinical trials show the anti-tumour potential of tumour-infiltrating lymphocytes and natural killer cells. More recently the success of CAR-T cells (Chimeric Antigen Receptors T cell) for haematological indications, the immune therapy space is currently exploring an array of different cellular immunotherapies for the treatment of cancer [
1‐
3]. Similar approaches also hold promise for non-cancer indications such as autoimmune diseases [
2]. However, as infused cells in contrast to non-living drugs in most cases have unknown in vivo distribution, the efficiency of cellular therapies cannot always be fully predicted. This is also true for side effects that are usually caused by the off-target accumulation of the infused cells in the healthy organs. One well-known example of this is graft-versus-host disease (GvHD) [
4]. Consequently, to ensure effective treatment while minimizing complications from off-target toxicity is essential to develop reliable methods that can dynamically determine the in vivo biodistribution of the infused cells [
5]. Long-term cell tracking with long-lived radionuclide-based tracers can provide the necessary information on cell behaviour and migration in vivo with real-time nuclear imaging [
6]. A vast number of radiotracers have been investigated for long-term cell tracking in vivo. The clinically used Single-Photon Emission Computerized Tomography (SPECT) radiotracer [
111In]In-oxine (also denoted as [
111In]In-(oxinate)
3) is suboptimal due to the limited spatial resolution of SPECT in vivo and isotope leakage [
7]. Current PET radiotracers are in many ways superior to SPECT regarding half-life, resolution and stability and might provide higher cellular retention. The isotope zirconium-89 (
89Zr) is an attractive isotope within PET imaging and fulfils several parameters required for cell tracking. Two of the most common chelators developed for
89Zr are oxine and DFO-NCS. The main difference between these two radiotracers is the cell labelling mechanism. The [
89Zr]Zr-(oxinate)
4 passively diffuses over the cell membrane where the complex dissolves. The 89Zr binds to unspecific molecules inside the cell, primarily in the cytosol, cell membrane, nucleus, chromatin and cytoskeleton [
8‐
10]. The [
89Zr]Zr-DFO-NCS binds to any free amine available on molecules on the cell membrane surface. Today, [
89Zr]Zr-(oxinate)
4 (also denoted as [
89Zr]Zr-oxine) is a well-evaluated radiotracer in preclinical studies and is currently in a first-in-human clinical study [
9,
11‐
15]. However, some limitations regarding radioactive leakage during the first 24 h should be further evaluated [
10,
16,
17]. Our group have recently optimized the synthesis and cell labelling of both [
89Zr]Zr-(oxinate)
4 and [
89Zr]Zr-DFO-NCS [
17]. Both radiotracers were successfully synthesized with a radiochemical yield (RCY) of > 95% and used to label different cell types with high labelling efficiency. This study aims to directly compare these two radiotracers in terms of in vivo biodistribution and preliminary dosimetry in rats.
Discussion
Even though cell labelling with PET radiotracers is a promising technique for long-term cell tracking, there is always a risk of affecting the cells. When labelling cells, it is seldom all cells that become radiolabelled, so the difficulty is to assess whether the radiolabelled cells behave the same as the unlabelled cells. The risk of an extracellular labelling method such as [
89Zr]Zr-DFO-NCS is that it could affect the cells´ interaction with surrounding tissues, hence altering the behaviour and in vivo distribution of the labelled cells. There are few such risks with an intracellular labelling method like [
89Zr]Zr-(oxinate)
4; on the other hand, it cannot be excluded that the use of this method could interfere with intercellular mechanisms and alter the cells’ functionality and expression. These risks have to be considered and evaluated individually for each cell type and radiotracer. The effects caused by radiolabelling on hDSC, rMac and PBMC with [
89Zr]Zr-(oxinate)
4 and [
89Zr]Zr-DFO-NCS were previously evaluated in vitro [
17]. There it is stated that there was no significant decrease in viability, proliferation and radioactive retention for any of the [
89Zr] Zr-(oxinate)
4 labelled cell lines 7 days post-labelling. The same results were seen for [
89Zr]Zr-DFO-NCS labelled hDSC. Immune cells labelled with [
89Zr]Zr-DFO-NCS did not show any significant decrease in proliferation or viability. The radioactive retention for [
89Zr]Zr-DFO-NCS labelled immune cells was significantly lower than cells labelled with [
89Zr] Zr-(oxinate)
4, with − 55% and − 25% for rMac and PBMC at day 7, respectively [
17]. However, hDSC showed signs of cellular stress, while rMac showed a slight decrease in phagocytosis function.
In this study, we evaluated the short-term (up to 4 days) effects radiolabelling might have on rMac proliferation. For the [
89Zr]Zr-(oxinate)
4 labelled cells, radioactive retentions appeared stable from 24 h and no significant difference in cell count compared to controls after 4 days. Cells radiolabelled with [
89Zr]Zr-DFO-NCS showed a significant decrease in cellular proliferation compared to controls (
p = 0.03) with no increase in cell count at day 4. The radioactive retention for both [
89Zr]Zr-(oxinate)
4 and [
89Zr]Zr-DFO-NCS labelled cells showed an initial drop during the first 24 h. After which the radioactive loss appeared to stabilize for [
89Zr]Zr-(oxinate)
4, while [
89Zr]Zr-DFO-NCS slowly continued to decrease until day 4. The loss in retention from [
89Zr]Zr-DFO-NCS is similar to what was observed in our previous study; the drop in retention during the first 24 h is likely due to the loss of cells caused by the radiolabelling procedure. The drop in retention for [
89Zr]Zr-(oxinate)
4 is likely due to leakage since there was no significant cell loss. Even though in the previous study the cell dose exceeded the recommended limit for risk of DNA damage, limited damage was detected after 7 days [
17]. Here we corrected the radioactive dose between the radiotracers, due to the larger number of cells the dose (MBq/10
6) to rMac was 1–2 times lower compared to hDSC. Henceforth, the only difference between the same cell line is the use of radiotracer; therefore, we can assume that neither the radioactive dose nor the labelling procedure is the cause of the accumulation in the lungs.
In this study, when comparing [
89Zr]Zr-(oxinate)
4 and [
89Zr]Zr-DFO-NCS, with both rMac and hDSC, we observed different migration patterns of the cells depending on the radiotracer used. Initially, both [
89Zr]Zr-(oxinate)
4 and [
89Zr]Zr-DFO-NCS show a rapid accumulation in the lungs. A large part of the signal from [
89Zr]Zr-DFO-NCS labelled cells stay in the lung until day 7, while [
89Zr]Zr-(oxinate)
4 labelled cells rapidly follow the expected pattern and continue migration to the liver [
10,
14,
15,
19,
24].
The control rats injected with unbound [89Zr]Zr-(oxinate)4 demonstrate a similar biodistribution pattern as the [89Zr]Zr-(oxinate)4 labelled cells, with high uptake in the spleen and liver. This can complicate the confirmation of the cells’ location without invasive biopsies. We see an almost identical signal in the heart compartment for radiolabelled hDSC compared to the controls with unlabelled [89Zr]Zr-(oxinate)4. As for the rMac, there is a somewhat prolonged signal in the heart which can be due to the sensitivity of rMac resulting in a slightly larger degree of cell death. It is therefore highly important that the cells are in good condition upon injection since damaged cells will be degraded and the radiotracers might redistribute and confound biodistribution analyses; all injected cells in this study showed 82 ± 6.4% viability upon injection. In the radioactive distribution in control rats that received [89Zr]Zr-(oxinate)4 labelled cells, we do not see a significant uptake in bone. This indicates that the efflux from dead or damaged cells is not in the form of unbound 89Zr. It is more likely that the 89Zr is still bound to oxine or conjugated to unspecific structures inside the cell, which is then transported to the liver and spleen instead of bone.
When comparing the controls injected with unbound hydrolysed [89Zr]Zr-DFO-NCS with the [89Zr]Zr-DFO-NCS labelled cells, they have a different biodistribution. This indicates that it is the cells that have accumulated in the lungs and not just [89Zr]Zr-DFO-NCS released from the cell surface. Along with the results from the in vitro stability where 47 ± 14% of the radioactivity is still attached to the cells after 4 days, it is most likely that the majority of signals represent the cells' whereabouts in vivo. We do see a slightly faster lung clearance from rMac labelled with [89Zr]Zr-DFO-NCS compared to hDSC, and this could also be explained by a higher degree of cell death due to the sensitivity of rMac. Moreover, dead cells and cell fragments are known to be excreted through the spleen and liver. Since the signals in these organs are hardly detectable, it indicated that the majority of [89Zr]Zr-DFO-NCS labelled cells are alive in the lungs. The heart signal from [89Zr]Zr-DFO-NCS labelled cells shows a similar pattern as the [89Zr]Zr-(oxinate)4 labelled cells. These data are based on the ROI of the whole heart, defined without the presence of CT which can result in measurement errors, especially with the proximity to the lungs which can entail a false positive measurement in the heart. Any free 89Zr in the blood will rapidly accumulate in the bone; thus, the lack of bone uptake there is hardly any leakage of free 89Zr. With the 89Zr still bound to unspecific molecules and proteins, the majority of radioactive leakage from dying cells will end up in the liver, spleen and kidneys. The degree of radioactive leakage from [89Zr]Zr-DFO-NCS labelled cells would explain why the whole-body retention is lower for rats receiving [89Zr]Zr-DFO-NCS labelled cells compared to [89Zr]Zr-(oxinate)4. Both the loss in whole-body retention and the cellular efflux from radiolabelled rMac after 24 h is 50% higher for [89Zr]Zr-DFO-NCS compared to [89Zr]Zr-(oxinate)4.
When comparing the radioactive dosimetry from radiolabelled cells, the high accumulation of [89Zr]Zr-(oxinate)4 in both the liver and spleen is compensated by the [89Zr]Zr-DFO-NCS high uptake in the lungs. What increases the effective dose for [89Zr]Zr-DFO-NCS is the high lung signal which also irradiates surrounding tissues. The proximity of the lungs to the spine, ribcage and shoulders, causes a substantial radioactive crossfire from the lungs to the bone marrow. Although [89Zr]Zr-(oxinate)4 shows an overall higher %/IA/g uptake in the bone, the high signal from the lungs from [89Zr]Zr-DFO-NCS labelled cells increases irradiation of the bone marrow, henceforth the effective dose.
Although conflicting data have been reported, Basal et al. present similar data with high and prolonged uptake in the lungs, which corresponds to our findings [
6]. Basal et.al. use a DFO-NCS concentration almost four times higher than in our study. Compared with our study, the only substantial difference between these reports is the concentration of DFO-NCS, while in another study by Lee et. al, they used roughly 800 times lower [
89Zr]Zr-DFO-NCS concentration and they reported a rapid migration to the liver [
15]. It is plausible that a high concentration of [
89Zr]Zr-DFO-NCS substantially blocks and disrupts essential surface receptors on the cells. If these surface structures are needed for tissue interaction, this might therefore hinder the labelled cells to migrate from the lungs to the liver.
Attempts to lower the concentration and label cells have been proven problematic due to the high loss of [89Zr]Zr-DFO-NCS and substantial handling of radioactivity. Since it is difficult to synthesise a stock solution of [89Zr]Zr-DFO-NCS with the required concentration and with high specific activity, a larger batch of [89Zr]Zr-DFO-NCS is required. If we only take 2% of the [89Zr]Zr-DFO-NCS stock solution, we match Lee et al. concentrations of 8.0 pmol DFO-NCS per 5 × 106 cells. The alternative was to increase the number of cells which was not feasible due to the limited harvest per donor. Still, the conflicting reports on the [89Zr]Zr-DFO-NCS labelled cells’ behaviour in vivo prove the need for further studies. If this is to be a reliable labelling technique for long-term cell tracking in vivo, we have to ensure that the radiolabelled cells mimic the behaviour of the unlabelled cells set for therapy.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.