Pharmacokinetic and pharmacodynamic studies of CD19 CAR T cell in human leukaemic xenograft models with dual‐modality imaging

Abstract In recent years, chimeric antigen receptor T (CAR T)‐cell therapy has shown great potential in treating haematologic disease, but no breakthrough has been achieved in solid tumours. In order to clarify the antitumour mechanism of CAR T cell in solid tumours, the pharmacokinetic (PK) and pharmacodynamic (PD) investigations of CD19 CAR T cell were performed in human leukaemic xenograft mouse models. For PK investigation, we radiolabelled CD19 CAR T cell with 89Zr and used PET imaging in the CD19‐positive and the CD19‐negative K562‐luc animal models. For PD evaluation, optical imaging, tumour volume measurement and DNA copy‐number detection were performed. Unfortunately, the qPCR results of the DNA copy number in the blood were below the detection limit. The tumour‐specific uptake was higher in the CD19‐positive model than in the CD19‐negative model, and this was consistent with the PD results. The preliminary PK and PD studies of CD19 CAR T cell in solid tumours are instructive. Considering the less efficiency of CAR T‐cell therapy of solid tumours with the limited number of CAR T cells entering the interior of solid tumours, this study is suggestive for the subsequent CAR T‐cell design and evaluation of solid tumour therapy.


| INTRODUC TI ON
Currently, there are 468 targets involved in the development of 4,000 immunotherapy drugs worldwide. 1 Among them, CD19 targets are always at the top of the research hot spots (192), and drugs targeting CD19 are already on the market. 2 CD19 is a type of leucocyte differentiation antigen, an important membrane antigen related to B-cell proliferation, differentiation, activation and antibody production, and is the best marker for diagnosing B-cell lineage tumours (leukaemia and lymphoma) and identifying B cells. 3 Chimeric antigen receptor (CAR) T therapy is a new type of cellular therapy technology, with special clinical results against haematoma. 4 At present, about half of the CAR T-cell targets are CD19, and the two CD19 CAR T products, Kymriah 5 and Yescarta, 6 have been approved for marketing by FDA (Food and Drug Administration) and EMA (European Medicines Agency). CD19 is the most frequent antigen target in the CAR T clinical projects accepted by the National Medical Products Administration (NMPA) in China. 7 Currently, CAR T cells have made some breakthroughs in the treatment of haematological tumours, but reports of solid tumours targeting CD19 are still rare.
CAR T cells are effective in treating haematologic malignancy directly in contact with blood tumour cells. 8 The efficacy of CAR T therapy is closely related to the number and activity of CAR T cells for the haematoma patients. While for the solid tumours, CAR T cells need to be transferred from peripheral blood to interior solid tumour and traverse the solid masses formed in the early stage to exert antitumour effects, the number and activity of CAR T cells migrating to solid tumour tissues are greatly reduced. 9 In addition, another major challenge is the selection of target because tumour tissue is heterogeneous and multiple target antigens may be expressed in the same tumour. 10 Moreover, there are also some adverse effects during the CAR T-cell therapy. 11 To meet these challenges, there is an urgent need to develop a sensitive, real-time, continuous, comprehensive and accurate visualization technique to reveal the distribution, migration and targeting of CD19 CAR T cells in vivo. 12,13 Current clinical monitoring of the efficacy of CAR T therapy includes tumour biopsy and DNA copy number in blood. 14,15 Tissue biopsies are difficult for some patients, and the sample may not represent accurate result due to the heterogeneity within the solid tumour. 16 DNA copy number in blood is currently not very sensitive, and the indicators in blood and bone marrow do not correlate with the infiltration of tumour masses. 17,18 The lack of monitoring methods of CAR T cells' biological behaviour in vivo has greatly limited its development and application.
Molecular imaging techniques, such as magnetic resonance imaging (MRI), single-photon emission computerized tomography (SPECT), positron emission tomography (PET) and fluorescent imaging, which can provide non-invasive, reproducible and quantitative tracking of implanted cells, might elucidate the above problems. 19 After labelled with superparamagnetic iron oxide nanoparticle (SPIO) agents, MRI can be used to monitor the SPIO-labelled stem cells from the injection site to the infarct area. 20 However, it is difficult to achieve whole-body imaging of the distribution of SPIOlabelled cells by MRI, as the dark signal induced by SPIOs may also be derived from other sources. 21 111 In-oxine labelling is always a gold standard approach for cell tracking in vivo by scintigraphy or SPECT.
However, this method requires relatively high levels of radioactivity, which might induce cellular damage. 22 In comparison with SPECT, PET has higher resolution and higher sensitivity, 23  Here, in order to verify the targeting and effectiveness of CD19 CAR T cells in CD19 high expression leukaemic solid tumours, we used 89 Zr as the radionuclide to label CAR T cells, and then, PET and optical imaging were used to investigate the in vivo PK and PD of CAR T cells.

| Targeting of CD19 CAR T cells
CD19 CAR T cells were generated as described by LiQing Kang et al. 18 To verify in vitro targeting validation of CD19 CAR T cells, we conducted cytotoxicity assays using the Cytotoxicity Detection Cell proliferation assays were performed using a carboxyfluorescein diacetate succinimidyl ester (CFSE) assay kit (Abcam, Cambridge, UK) following the manufacturer's instructions. In brief, the CAR T cells were labelled with 2.5 μM CFSE and then co-cultured with CD19-positive/negative cells, which treated with mitomycin before to stop the division, at a stimulator-to-responder ratio of 5:1 (10 6 CAR T cells/mL) for 5 days in 24-well plates in 500 μL serum-free AIMV (Gibco) medium per well. Flow cytometry was performed using an Attune NxT flow cytometer (Thermo Fisher) to detect changes in CFSE intensity. FlowJo v10 software (TreeStar, San Carlos, CA, USA) was used for data analysis.

| Radiolabelling of CAR T cells
CD19 CAR T cells were radiolabelled with 89 Zr using methods reported before with some modifications. 27

| Group and treatment
The mice were then randomly divided into four groups as shown in Table 1. Group 1 (n = 12) and group 3 (n = 4) were the CD19-positive xenograft model. Group 2 (n = 12) and group 4 (n = 3) were the CD19-negative xenograft model. Group 1 and group 2 were treated with 2*10 6 89 Zr-labelled CAR T via tail vein. Group 3 and group 4 were treated with saline.

| Pharmacodynamic study
To evaluate the therapeutic efficacy of CD19 CAR T cells in human leukaemic xenograft models, CD19-positive model and CD19-negative model mice were employed and received CD19 CAR T therapy. The tumour volume and bodyweight were monitored every two days for 25 days, as well as the state and survival of the animal models.  and data were analysed using software that comes with the instrument.

| micro-PET Image
In order to monitor the in vivo distribution of CD19 CAR T cells, The image-derived percentage of injected dose per gram (%ID/g) was calculated for each ROI and used as the indicator for the quantification of radioactivity uptake. The %ID/g can be obtained using the following equations: In order to obtain the kinetics of 89 Zr-CAR T-cell distribution in vivo, we mapped the uptake curve of four main organs (lung, liver, spleen and tumour) with representative mean %ID/g.

| Ex vivo biodistribution and blood pharmacokinetics
Six CD19-positive model and six CD19-negative model mice were PerkinElmer). The uptake (%ID/g) of blood was then calculated by the following equation: Pharmacokinetic parameters, such as AUC and half-time (t 1/2 ), were then fitted with DAS (version 1.1) software. 28 After killing the mice on day 7, blood and organs (brain, heart, liver, spleen, lung, kidney, stomach, duodenum, colon, muscle, sexual gland, fat, tibia, joint, marrow, adrenal gland, gladder and tumour) were collected, weighed immediately and γ-counted together with standards prepared from a sample of injected material. The percentage of injected dose per gram (% ID/g) of tissue was calculated by the following equation:

| Specific tumour uptake
Referring to the previous study, 29  In view of the strong correlation between blood samples taken from tail and heart through PET imaging, 29 and moreover, the data from tail blood are more accurate and reliable than heart blood in this study, the blood activities from tail were used as the input functions to establish the Patlak model. In addition, the B-value was used as a key parameter for the tumour-specific uptake of CD19 CAR T cells in CD19-positive tumours.

| Statistical analysis
Results are expressed as the mean ± SD unless stated otherwise.
Statistical comparisons between two groups were evaluated with

| In vitro targeting validation of CD19 CAR T cells
To subgroups of CAR T cells, the peak plot of the K562-CD19 group showed a significant left shift, suggesting that CAR T cells proliferated, while the T cells did not. Furthermore, fluorescent intensity peak plots of its three subgroups were similar, reflecting the cells did not proliferate significantly. In addition, the K562-CD19 peak plots of NC and CAR T cells were superimposed in Figure 1D. The fluorescence of CFSE was highly attenuated, which indicates that CD19-K562-luc cells can specifically stimulate CD19 CAR T cells to induce proliferation.

| Radiolabelling of CAR T cells
Consistent with the results of our previous study, 27 the cell viability (>90%) and 89 Zr retention (>80%) of the CAR T cells changed slightly after radiolabelling with 89 Zr-oxine for 48 hours. In addition, the radiolabelling yield was promoted from 10% to 40% due to the increase in the amount of CAR T cells. Radiochemical purity was >95%.

| Pharmacodynamic study
The results of tumour cell activity evaluated with bioluminescence imaging are shown in Figure 2A A survival study was performed as well ( Figure 3D), and the results indicated that the CD19-positive group died from day 17 until day 30 due to individual differences in therapy efficacy. In contrast, the CD19negative model mice died around day 15 due to ineffective treatment and adverse reactions. In addition, the two treated groups achieved bodyweight loss during 0-15 days, and then, they gradually restored, while the bodyweight of the two control groups did not decrease during the whole experimental period ( Figure 3E).

| In Vivo PET Tracking of 89 Zr-Radiolabelled CAR T cells
To further validate the tumour-targeting performance of 89 Zr-labelled CD19 CAR T on CD19-positive tumour in vivo, micro-PET imaging was performed in the CD19-positive group and CD19-negative group.
Tomographic images of multiple time-points after intravenous injection are shown in Figure 4A. The uptake (%ID/g) of tumour, lung, liver and spleen is shown in Figure 4B. Consistent with the results of our previous study, 27 in mice of two groups, 89 Zr-labelled CAR T cells were distributed in the lungs at 2 hours (71.36 ± 16.27%ID/g for the CD19positive and 79.74 ± 6.19%ID/g for the CD19-negative group) and then rapidly decreased to 1.37 ± 0.29%ID/g (CD19-positive group) and 1.60 ± 0.20%ID/g (CD19-negative group) for the next 7 days.
The 89 Zr-CAR T cells were distributed in the liver (45.63 ± 3.54%ID/g for the CD19-positive and 45.75 ± 2.35%ID/g for the CD19-negative group) and spleen (46.07 ± 10.43%ID/g for the CD19-positive and 49.65 ± 7.67%ID/g for the CD10-negative group) at 24h and maintaining a plateau ( Figure 4C, D, E). No significant changes appeared in tumour uptake during the long-term monitoring, and on day 7, the uptake of tumour in CD19-positive tumour and CD19-negative tumour was 0.39 ± 0.07%ID/g and 0.40 ± 0.10%ID/g, respectively (P > .05). Figure 5, which revealed a high concentration of radiolabelled CAR T cells in spleen (590.58 ± 49.36%ID/g, n = 6) and liver ( 57.43 ± 11.86%ID/g, n = 6), followed by joint (13.84 ± 5.44%ID/g, n = 6) and marrow (11.70 ± 8.93%ID/g, n = 6) in the CD19-positive group, and the same is observed in the CD19-negative group, with spleen (452.48 ± 253.79%ID/g, n = 6), liver (70.94 ± 8.36%ID/g, n = 6), joint (16.08 ± 3.88%ID/g, n = 6) and marrow (27.79 ± 9.88%ID/g, n = 6). The tumour uptake (%ID/g) in the CD19positive group was 0.45 ± 0.28%ID/g and 0.32 ± 0.16%ID/g in the CD19-negative group. Uptake of 89 Zr-CAR T cells in tumours showed no major differences between the two groups. Compared with uptake calculated by PET image, there is no significant difference between all the organs except spleen. We speculate that this is because the spleen is so small that the uptake values of PET imaging were susceptible to boundary volume effects.

| Pharmacokinetics
As shown in Figure 6A, the curve of blood uptake of 89 Zr-labelled CD19 CAR T cells in two groups was fitted with DAS. At 5 minutes after injection, the blood uptake was 16.59 ± 6.46%ID/g in Currently, we can monitor the clinical efficacy of CAR T therapies by performing direct peripheral blood tumour-specific T-cell counts, serum analysis of cytokines associated with T-cell activation and (repeat) tumour biopsies. In this experiment, we also tried to use the qPCR method to measure DNA copy number in blood samples.
This method has two drawbacks: first, invasive collection of at least 100 µL of blood for qPCR analysis is difficult to consecutively obtain in small animal models, while PET is a non-invasive and real-time dynamic monitoring; second, the distribution of CAR T cells in all normal tissues throughout the body cannot be examined to assess their safety; and furthermore, in this study, the test results of qPCR were below the detection limit regardless of the efficacy, indicating that the sensitivity of this method might be inferior to PET.
In addition to the difficulty of CAR T cells to migrate inside solid tumours, another major challenge to the efficacy of solid tumours is the selection of targets. 30 The effectiveness of CAR T therapy is closely related to targeting. Our preliminary PK and PD results suggest that PET imaging using 89 Zr-labelled CAR T cells can be used for the targeting validation of CAR T cells, and is expected to be used for subsequent CAR T-cell target screening and efficacy prediction.
Nevertheless, the cell tracking method in this study has certain limitations: on the one hand, direct labelling CAR T cells can only reveal the distribution, migration and homing of parental cells in vivo, and cannot show the cell proliferation, activation or death; on the other hand, we did not further validate this finding in a Raji xenograft model. The selection of animal models is particularly important when conducting CAR T therapy PK/PD studies as a suitable animal model can mimic the human in vivo tumour microenvironment and reflect drug targeting on CAR T-cell mechanism of action.
Subsequently, we plan to further validate this PK/PD result in situ haematoma animal models.
In summary, we used PET and optical imaging to investigate the PK and PD of CD19 CAR T-cell target therapy for leukaemic solid tumour. CD19-positive solid tumours had specific targeting uptake and better tumour suppression efficiency after CD19 CAR T treatment.
This approach is promising to provide a basis for PK-PD investigation

CO N FLI C T O F I NTE R E S T
The authors declared no potential conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.