Here, we describe a protocol in which an acute lymphoblastic leukemia patient-derived xenograft model is used as a strategy to assess and monitor CD19-targeted chimeric antigen receptor T cell-associated toxicities.
Chimeric antigen receptor T (CART) cell therapy has emerged as a powerful tool for the treatment of multiple types of CD19+ malignancies, which has led to the recent FDA approval of several CD19-targeted CART (CART19) cell therapies. However, CART cell therapy is associated with a unique set of toxicities that carry their own morbidity and mortality. This includes cytokine release syndrome (CRS) and neuroinflammation (NI). The use of preclinical mouse models has been crucial in the research and development of CART technology for assessing both CART efficacy and CART toxicity. The available preclinical models to test this adoptive cellular immunotherapy include syngeneic, xenograft, transgenic, and humanized mouse models. There is no single model that seamlessly mirrors the human immune system, and each model has strengths and weaknesses. This methods paper aims to describe a patient-derived xenograft model using leukemic blasts from patients with acute lymphoblastic leukemia as a strategy to assess CART19-associated toxicities, CRS, and NI. This model has been shown to recapitulate CART19-associated toxicities as well as therapeutic efficacy as seen in the clinic.
Chimeric antigen receptor T (CART) cell therapy has revolutionized the field of cancer immunotherapy. It has proven to be successful in treating relapsed/refractory acute lymphoblastic leukemia (ALL), large B cell lymphoma, mantle cell lymphoma, follicular lymphoma, and multiple myeloma1,2,3,4,5,6,7, leading to recent FDA approvals. Despite the initial success in clinical trials, treatment with CART cell therapy results in toxicities that are often severe and occasionally lethal. The most common toxicities after CART cell therapy include the development of CRS and NI, also referred to as immune effector cell-associated neurotoxicity syndrome (ICANS)8,9. CRS is caused due to the overactivation and massive expansion of CART cells in vivo, leading to the subsequent secretion of multiple inflammatory cytokines, including interferon-γ, tumor necrosis factor-α, granulocyte-macrophage colony-stimulating factor (GM-CSF), and interleukin-6 (IL-6). This results in hypotension, high fevers, capillary leak syndrome, respiratory failure, multi-organ failure, and in some cases, death10,11. CRS develops in 50-100% of cases after CART19 cell therapy11,12,13. ICANS is another unique adverse event associated with CART cell therapy and is characterized by generalized cerebral edema, confusion, obtundation, aphasia, motor weakness, and occasionally, seizures9,14. Any grade of ICANS occurs in up to 70% of patients, and Grades 3-4 are reported in 20-30% of patients5,10,15,16. Overall, CRS and ICANS are common and can be fatal.
The management of ICANS after CART cell therapy is challenging. Most patients with ICANS also experience CRS17, which can often be treated with the IL-6 receptor antagonist tocilizumab or steroids18. A previous report revealed that early intervention with tocilizumab decreased the rate of severe CRS but did not affect the incidence or severity of ICANS19. Currently, there is no effective treatment or prophylactic agent for ICANS, and it is crucial to investigate preventive strategies20.
Myeloid cells and associated cytokines/chemokines are thought to be the main drivers of the development of CRS and ICANS21. While CRS is directly related to the extreme elevation of cytokines and T cell expansion, the pathophysiology of ICANS is largely unknown22,23. Therefore, it is imperative to establish a mouse model that recapitulates these toxicities after CART cell therapy to study the mechanisms and develop preventive strategies.
There are multiple preclinical animal models currently used to study, optimize, and validate the efficacy of CART cells, as well as to monitor their associated toxicities. These include syngeneic, xenograft, immunocompetent transgenic, humanized transgenic, and patient-derived xenograft mice, in addition to primate models. However, each of these models has drawbacks, and some do not reflect the true efficacy or safety concerns of CART cells24,25. Therefore, it is imperative to carefully choose the best model for the intended goals of the study.
This article seeks to describe the methodology that is used to assess CART cell-associated toxicities, CRS and NI, using an ALL patient-derived xenograft (PDX) in vivo model (Figure 1). Specifically, in the methods described here, CART19 cells generated in the authors' laboratory are used following previously described protocols. Briefly, human T cells are isolated from healthy donor peripheral blood mononuclear cells (PBMCs) via a density gradient technique, stimulated with CD3/CD28 beads on day 0, and lentivirally transduced on day 1 with CARs composed of a CD19-targeted single chain variable fragment fused to 4-1BB and CD3ζ signaling domains. These CART cells are then expanded, de-beaded on day 6, and cryopreserved on day 826,27,28,29,30. As outlined previously, mice are subjected to lymphodepleting treatment, followed by the administration of patient-derived leukemic blasts (ALL)28. First, tumor engraftment is verified via submandibular blood collection. Following the establishment of an appropriate tumor burden, CART19 cells are administered to the mice. Then, the mice are weighed daily to assess well-being. Small animal magnetic resonance imaging (MRI) is performed to assess NI, along with tail bleeding to assess T cell expansion and cytokine/chemokine production. The techniques described below are highly recommended to be used as a model to study CART cell-associated toxicities in a PDX model.
This protocol follows the guidelines of Mayo Clinic's Institutional Review Board (IRB), Institutional Animal Care and Use Committee (IACUC A00001767), and Institutional Biosafety Committee (IBC, Bios00000006.04).
NOTE: All the materials used to work with mice must be sterile.
1. Injection of busulfan to NSG mice
2. Injection of ALL patient-derived blasts (CD19+) to the NSG mice
NOTE: This protocol follows the guidelines of Mayo Clinic's Institutional Biosafety Committee (IBC, Bios00000006.04).
3. Tumor engraftment assessment
4. Administration of CART19 cells in vivo
5. Assessment of CART19 cell-associated toxicities
6. MRI imaging
NOTE: A preclinical (for small animals) MRI scanner with a vertical bore magnet was used for in vivo magnetic resonance and image acquisition32,33.
The aim of this protocol is to assess CART cell-associated toxicities using a PDX mice model from tumor cells of patients with ALL (Figure 1). First, NSG mice received i.p. injections of busulfan (30 mg/kg) with the goal of immunosuppressing them and facilitating CART cell engraftment28. The following day, they received ~5 × 106 PBMCs (i.v.) derived from ALL patients. The mice were monitored for engraftment for ~13 weeks via the tail bleeding assay. Red blood cells were collected and lysed, followed by flow cytometry preparation to assess the expression of CD19+ tumor cells (Figure 2).
Once the mice reached ≥10 tumor cells/µL, they were randomized and prepared to receive 4 × 106 CART19 cells (i.v.). To assess CART19-associated toxicities, the mice were weighed once per day (Figure 3A). Weight loss was chosen as a symptom of CRS appearance, as it has been previously revealed to be related to CART cell-associated toxicities28,37. Peripheral blood assays (via submandibular blood collection) were performed to assess tumor burden and CART cell expansion by flow cytometry analysis of human CD19+ tumor cells and human CD3+ T cells. In addition, serum was collected, and an assessment of inflammatory cytokines was performed. Assays were done one day prior to CART19 cell infusion as well as at 5 and 10 days after CART19 cell infusion, as these timepoints are particularly relevant to CRS onset and development of symptoms.
Here, in particular, a Multiplex assay was performed before and after CART19 cell administration, where the most representative cytokines and chemokines were GM-CSF, IL-18, MIP-1α, and IP-10 (Figure 3B). However, other assays can be done to assess the concentrations of these proteins in the serum. Alternatively, mice were subjected to MRI scanning to evaluate vascular permeability in the central nervous system (CNS). Mice were injected with the contrast reagent gadolinium (i.p.) to measure brain inflammation and blood-brain barrier disruption38. Then, they were anesthetized with isoflurane, and contrast-enhanced T1-weighted and T2-weighted images were taken (Figure 4). With T2 images (Figure 4, left), the brighter section (white) indicates the presence of edema or fluid, which is mostly seen in brain tumors and inflammation. With T1 images (Figure 4, middle), the brighter sections indicate areas where there is blood leaking or blood-brain barrier disruption. Lastly, with the help of software, the 3D reconstruction images (Figure 4, right) were assembled using gadolinium-enhanced T1 images based on hyperintensity signals corresponding to vascular permeability, which renders the volume of gadolinium leakage in the brain34.
Figure 1: Experimental scheme of a patient-derived xenograft model used to assess CART cell-associated toxicities, cytokine release syndrome, and neuroinflammation. NSG mice were conditioned with 30 mg/kg busulfan (i.p.) and received 3 × 106-5 × 106 blasts (i.v.) derived from the peripheral blood of patients with ALL. Next, the mice were monitored for tumor engraftment via submandibular blood collection, followed by flow cytometry to assess CD19+ blasts. When the peripheral blood CD19+ cells were ≥10 cells/µL, the mice received 2 × 106-5 × 106 CART19 cells. The mice were weighed daily as a measure of their well-being. At 10 days post CART19 cell infusion, mouse brain MRIs were performed to detect neuroinflammation. Tail bleeding to assess cytokine/chemokine production and T cell expansion was performed weekly post CART19 cell injection. Abbreviations: CART = chimeric antigen receptor T cell; CART19 = CD19-targeted CART; ALL = acute lymphoblastic leukemia; i.v. = intravenous; i.p. = intraperitoneal; PB = peripheral blood; NSG = NOD-SCID IL2rγnull. Please click here to view a larger version of this figure.
Figure 2: Analysis of CD19+ tumor cells as an indicator of tumor burden in an ALL PDX model to assess CART-cell associated toxicities. (A) Representative flow cytometric analysis showing the gating strategy to analyze the CD19+ cell population taken from a peripheral blood sample of a mouse injected with patient-derived ALL blasts. (B) The quantification and comparison of the CD19+ cell populations between mice treated with CART19 versus untreated mice based on raw data obtained from flow cytometry analysis. (one-way ANOVA; n = 3 mice per group; error bars, SEM). Abbreviations: ns = not significant; ALL = acute lymphoblastic leukemia; PDX = patient-derived xenograft; CART = chimeric antigen receptor T cell; CART19 = CD19-targeted CART; SSC-H = side scatter-peak height; FSC-H = forward scatter-peak height; SSC-A = side scatter-peak area; FSC-A = forward scatter-peak area. Please click here to view a larger version of this figure.
Figure 3: Assessment of CART19-associated toxicities using an ALL PDX model. (A) Representative graph of the percent weight reduction in mice treated with CART19 cells or control untransduced T cells for the first 6 days after CART (two-way ANOVA, ***p < 0.001, ****p < 0.00001; error bars, SEM). (B) Expression of cytokines and chemokines in NSG mice serum before and after CART19 cell administration. NSG mice were subjected to submandibular bleeding before and after CART19 cell administration (i.v.) to collect their serum and assess the expression of the following cytokines and chemokines: GM-CSF, IL-18, MIP-1α, IP-10, IL-1β and IL-6 (two-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; n =3 mice; nine technical replicates; error bars, SEM). Abbreviations: ns = not significant; ALL = acute lymphoblastic leukemia; PDX = patient-derived xenograft; CART = chimeric antigen receptor T cell; CART19 = CD19-targeted CART; NSG = NOD-SCID IL2rγnull; GM-CSF = granulocyte-macrophage colony-stimulating factor; IL-18 = interleukin-18; MIP-1α = macrophage inflammatory protein-1 alpha; IP-10 = interferon gamma-induced protein 10; IL-1β = interleukin-1β; IL-6 = interleukin-6. Please click here to view a larger version of this figure.
Figure 4: Magnetic resonance imaging used to assess neuroinflammation during CART19 cell treatment in a PDX model. NSG mice engrafted with tumor cells derived from ALL patients were infused with CART19 cells, and the brains were imaged with gadolinium-enhanced MRI after the mice exhibited neurological symptoms within 10 days after CART infusion. (Left) T2-weighted image revealing evidence of edema and possible inflammatory infiltrate during NI in CART19-treated mice (red arrow). (Middle) Representative axial section of T1-weighted MRI showing contrast enhancement within the brain parenchyma, indicating increased vascular permeability in CART19-treated mice (red arrow). (Right) Three-dimensional reconstructions of the rodent brain with T1-hyperintensity regions (red) were generated using Analyze 12.0 software. Abbreviations: MRI = magnetic resonance imaging; ALL = acute lymphoblastic leukemia; PDX = patient-derived xenograft; CART = chimeric antigen receptor T cell; CART19 = CD19-targeted CART. Please click here to view a larger version of this figure.
Human | Mouse | |
Cytokine/Chemokine | Human myeloid cytokines (IL-6, GM-CSF, IL-1, MCP-1) | Human myeloid cytokines (IL-6, GM-CSF, IL-1, MCP-1) |
Time to CRS (Day) | 2-5 | 4-6 |
Time to NI (Day) | 5-7 | 6-8 |
BBB status | BBB disruption | BBB disruption |
Tissue resident macrophage | Microglial activation | Microglial activation |
CART cell in CNS | Brain infiltration | Brain infiltration |
Table 1: Recapitulation of the clinically observed CART cell-associated toxicities by the ALL PDX model. Abbreviations: GM-CSF = granulocyte-macrophage colony stimulating factor; IL = interleukin; CRS = cytokine release syndrome; NI = neuroinflammation; BBB = blood-brain barrier; CNS = central nervous system; MCP-1 = monocyte chemoattractant protein 1.
In this report, a methodology to assess CART cell-associated toxicities using an ALL PDX model has been described. More specifically, this model seeks to mimic two life-threatening toxicities, CRS and NI, that patients often experience after the infusion of CART cells. It recapitulates many hallmarks of CART toxicities observed in the clinic: weight loss, motor dysfunction, neuroinflammation, inflammatory cytokine and chemokine production, and the infiltration of different effector cells into the central nervous system8,20,28. Previously, this PDX model has been used to mimic the initial development of CRS and NI (Table 1) and assess different strategies for toxicity prevention and treatment. One study in this PDX model utilized CART19 cells depleted of GM-CSF. More specifically, these results demonstrated that GM-CSF depletion resulted in the reduced secretion of CRS-linked cytokines and chemokines, which was clearly demonstrated using this ALL PDX mouse model29. This PDX model has also been used successfully to assess CART cell in vivo expansion correlating with CRS development after CART19 cell infusion using a sodium iodide symporter (NIS) as an imaging platform26. Therefore, this proposed model represents a useful tool that can be used to assess the potential toxicities arising from CART cells.
It is important to mention that many other models have been used to study CART cells in vivo. These include syngeneic, transgenic, and humanized mouse models, in addition to xenogeneic models24. The syngeneic mouse model, also called the immunocompetent allograft mouse model, utilizes CART cells, tumor, and target antigens that are murine in origin. This is advantageous as it allows for the study of CART cells in a fully functional immune system and can also reveal on-target, off-tumor toxicities. The disadvantage of this model is that mouse biology and immunity are different from those of humans, and, therefore, it is not a relevant model for translational studies where the goal is to proceed to clinical trials24,39.
Immunocompetent transgenic mouse models express a human tumor-associated antigen (TAA) transgene on both healthy tissue and tumors in patterns mimicking those found in human tissues. These mice have also been used in CART studies to better evaluate the safety of CART cells by revealing on-target off-tumor effects40,41. Humanized mice implanted with CD34+ human hematopoietic stem cells can be an advantageous model to study CART efficacy and toxicity on the human hematopoietic system, as the reconstituted immune system can mimic CART inhibition and induce a cytokine storm that recapitulates CRS seen in the clinic42,43,44. Although these models allow the study of specific immune cells in relation to CART efficacy, the humanized system is still primitive and needs further optimization. In addition, their use in studying off-target effects is limited since most of the host antigens on healthy tissue are murine in origin. Furthermore, the CD34+ humanized mouse model is technically challenging to develop45. However, human xenograft models in immunocompromised mice are commonly used to assess human CART cell efficacy prior to translating to clinical trials, and as demonstrated in this paper, they can be used to evaluate the toxicities associated with CART cell activity. However, the lack of immune system and immune cells such as T cells and NK cells limits their application for assessing off-tumor on-target toxicity, interactions of CART cells with innate immune cells, and CART cell inhibition by the tumor microenvironment.
The advantage of the protocol described here is that it requires simple procedures using NSG mice to observe and assess CART cell-associated toxicities. First, it is easy to reproduce, and it has been optimized to assess strategies to prevent CART toxicities through GM-CSF neutralization and to visualize CART cell expansion in correlation with CRS onset. When using this novel PDX model, it is important to correctly perform the critical steps, such as confirming the proper engraftment of the ALL blasts and the early monitoring of CRS onset right after CART administration. The assessment of cytokines and weight monitoring are crucial for determining CART-associated toxicities, and MRI defines the onset of NI. The downsides of this model are as follows: 1) as PDX ALL cells take longer to engraft and reach a high tumor burden, it requires approximately 2-3 months from tumor cell inoculation to CART19 cell treatment; 2) as a primary PDX model, there is significant patient to patient variability with regard to the time to engraftment and response to CART cell therapy; 3) tumor load cannot be assessed with bioluminescence imaging and requires frequent peripheral blood assessment; and 4) the lack of innate immune cells limits the ability of this model to study CART interactions with immune cells or the tumor microenvironment.
Interest in testing adoptive cells within more complex mouse models has grown over the past decade as it has become evident that simple immunodeficient mouse models are not ideal preclinical models. The PDX mouse model described here to assess CART cell-associated toxicities represents a promising preclinical model in the field of CART cell therapy as it mimics the timeline, symptoms, and markers of CRS and NI after CART cell infusion as seen in the clinic. Overall, the methodology described here provides a robust platform for assessing CART19 cell-associated toxicities as well as efficacy.
The authors have nothing to disclose.
This work was partly supported through the National Institutes of Health (R37CA266344, 1K99CA273304), Department of Defense (CA201127), Mayo Clinic K2R pipeline (S.S.K.), the Mayo Clinic Center for Individualized Medicine (S.S.K.), and the Predolin Foundation (R.L.S.). In addition, we would like to thank the Mayo Clinic NMR Core Facility staff. Figure 1 was created in BioRender.com
APC Anti-Human CD19 | Biolegend | 302211 | |
Alcohol Prep Pad | Wecol | 6818 | |
Analyze 14.0 software | AnalyzeDirect Inc. | N/A | https://analyzedirect.com/analyze14/ |
Artificial tears (Mineral oil and petrolatum) | Akorn | 17478-062-35 | Topical ophtalmic gel to prevent eye dryness |
BD FACS Lysing Solution | BD | 349202 | Red blood cells lysing buffer |
BD Micro-Fin IV insulin syringes | BD | 329461 | |
Brillian Violet 421 Anti-Human CD45 | Biolegend | 304032 | |
Bruker Avance II 7 Tesla | Bruker Biospin | N/A | MRI machine |
Busulfan (NSC-750) | Selleckchem | S1692 | |
CountBright absolute counting beads | Invitrogen | C36950 | |
CytoFLEX System B4-R2-V2 | Beckman Coulter | C10343 | flow cytometer |
Dulbecco's Phosphate-Buffered Saline | Gibco | 14190-144 | |
ERT Control/Gating Module | SA Instruments | Model 1030 | Small Animal Monitoring Respiratory and Gating System |
Fetal bovine serum | Millipore Sigma | F8067 | |
Hemocytometer | Bright-Line | Z359629-1EA | |
Human AB Serum; Male Donors; type AB; US | Corning | 35-060-CI | |
Isoflurane (Liquid) | Sigma-Aldrich | 792632 | |
LIVE/DEAD Fixable Aqua Dead Cell Stain Kit, for 405 nm excitation | Invitrogen | L34966 | |
Microvette 500 Lithium heparin | Sarstedt | 20.1345.100 | Blood collection tube |
MILLIPLEX Huma/Cytokine/Chemokine Magnetic Beads Panel | Millipore Sigma | HCYTMAG-60K-PX38 | Immunology Multiplex Assay to identify cytokines and chemokines |
Omniscan | Ge Healthcare Inc. | 0407-0690-10 | Gadolinium-based constrast agent |
Pd Anti-Mouse CD45 | Biolegend | 103106 | |
Penicillin-Streptomycin-Glutamine (100x), Liquid | Gibco | 10378-016 | |
Round Bottom Polysterene Test tube | Corning | 352008 | |
Sodium Azide, 5% (w/v) | Ricca Chemical | 7144.8-16 | |
Stainless Steel Surgical Blade | Bard-Parker | 371215 | |
X-VIVO 15 Serum-free Hematopoietic Cell Medium | Lonza | 04-418Q |