Corticosteroids ameliorate CAR T-cell-induced cytokine-release syndrome without inhibiting multiple myeloma treatment

Abstract

Background

Cytokine-release syndrome (CRS) is a common toxicity of chimeric antigen receptor (CAR) T cells. CRS is often treated with corticosteroids such as dexamethasone. Dexamethasone is also used to treat multiple myeloma. To model CRS after CAR T-cell treatment of multiple myeloma (MM), three cell types are required: monocyte-lineage cells, CAR T cells, and MM cells. Some cytokines important in CRS, including interleukin (IL)-6, are released mainly by monocyte-lineage cells.

Methods

We added cells of an acute monocytic leukemia cell line (THP-1) to co-cultures of anti-B-cell maturation antigen (BCMA) CAR T cells (CAR-BCMA) and BCMA⁺ target cells. Addition of THP-1 cells to the co-cultures led to increased levels of IL-6 and monocyte chemoattractant protein-1 (MCP-1) in culture supernatants. We developed a murine CRS model. This model included engraftment of THP-1 into NOD-scid common γ-chain-deficient mice to provide a source of some cytokines associated with CRS, including IL-6 and MCP-1. The murine model also included engraftment of the bioluminescent BCMA⁺ MM cell line MM.1S-veff-Luc and an infusion of CAR-BCMA. We treated CRS with dexamethasone or vehicle control.

Results

With this model, mice exhibited signs of CRS and had elevated serum cytokine levels after CAR T-cell infusion, and CAR-BCMA eliminated large burdens of MM.1S-veff-Luc. Dexamethasone administered 1, 3, and 5 days after CAR-BCMA ameliorated CRS. Dexamethasone was associated with faster elimination of MM burdens when either a dexamethasone-sensitive cell line (MM.1S-veff-Luc) or a dexamethasone-resistant cell line (MM.1R-veff-Luc) was used as the malignancy burden. Importantly, mice that received CAR-BCMA plus dexamethasone had higher levels of splenic CAR T cells when compared with mice that received CAR-BCMA without dexamethasone. When MM.1S-veff-Luc was treated, the median splenic CD3⁺CAR⁺ cell count for mice that received CAR-BCMA plus dexamethasone was 764 473 vs 327 888 for mice that received CAR-BCMA without dexamethasone (p=0.0021).

Among four patients who received anti-BCMA CAR T cells and corticosteroids on a clinical trial, CAR⁺ cell levels continued to increase after initiation of corticosteroids in all patients.

Conclusions

In summary, our results should encourage further clinical research to design corticosteroid regimens that optimize treatment of CAR T-cell toxicities while maintaining anti-malignancy activity.

Trial registration number

NCT03602612.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Chimeric antigen receptor (CAR) T-cell therapy often causes cytokine-release syndrome (CRS) and neurological toxicity.

• Corticosteroids, such as dexamethasone, are effective at treating CAR T-cell toxicity, but a more complete understanding of how corticosteroids affect antimalignancy activity is needed.

WHAT THIS STUDY ADDS

• Using a new murine CRS model, we assessed how dexamethasone affected anti-B-cell maturation antigen CAR T-cell treatment of multiple myeloma.

• Administering dexamethasone after CAR T-cell treatment accelerated multiple myeloma elimination and increased splenic CAR T-cell levels.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our results should encourage clinical research to find optimal corticosteroid doses and schedules that effectively treat CAR T-cell toxicities but do not inhibit antimalignancy activity.

Background

Cytokine release syndrome (CRS) is one of the major toxicities of chimeric antigen receptor (CAR) T-cell treatment; in addition, neurological toxicity occurs after CAR T-cell therapy.1,4 Extensive evidence from many clinical trials has shown an association between toxicity after CAR T-cell infusions and serum levels of a variety of cytokines including interferon-γ (IFNγ), interleukin (IL)-8, IL-6, and monocyte chemoattractant protein-1 (MCP-1).5,7 Clinical manifestations of CRS include fever, tachycardia, and hypotension.^{1 4}

IFNγ, granulocyte-macrophage colony-stimulating factor (GM-CSF), tumor necrosis factor-α (TNF-α), and other cytokines released by CAR T cells could cause toxicity directly, or they could cause secondary release of IL-6, IL-1, MCP-1, and other cytokines from myeloid cells such as monocytes and macrophages.8,10 Giavridis et al established a CRS model that used severe combined immunodeficiency (SCID)-beige mice.¹¹ This model recapitulated many CRS symptoms and showed important roles for macrophages, IL-6, nitric oxide, and IL-1 in CRS.¹¹ Norelli et al established a murine CRS model that used humanized mice.¹² The mice had myeloid cells, leukemia cells, and CAR T cells all derived from human cells.¹² The mice manifested CRS-like symptoms after infusion of CAR T cells, and the investigators demonstrated important roles for monocytes, IL-1, and IL-6 in CRS.¹²

Corticosteroids such as dexamethasone are commonly used to manage CRS that is resistant to the IL-6 receptor antagonist tocilizumab.^{1 3 4 13 14} Natural endogenous corticosteroids typically inhibit T cells, reducing cytokine production and naïve T-cell proliferation.^{15 16} Corticosteroids can also inhibit cytokine production by monocytes.¹⁷ Early studies indicated that corticosteroids might impair CAR T-cell antileukemia activity and inhibit CAR T-cell proliferation.^{13 14} In a retrospective study of lymphoma treatment with T cells expressing the anti-CD19 CAR axicabtagene ciloleucel, higher cumulative corticosteroid doses and early use of corticosteroids were associated with shorter progression-free survival and shorter overall survival.¹⁸ On the other hand, different clinical studies did not show that corticosteroids had an impact on complete remission rates when acute lymphoblastic leukemia was treated with anti-CD19 CAR T cells.¹⁹ In addition, there was an excellent 80% complete response rate in patients with diffuse large B-cell lymphoma receiving three daily toxicity prophylactic doses of dexamethasone starting on the day of axicabtagene ciloleucel infusion; some of these patients also received additional later doses of corticosteroids for toxicity treatment.²⁰

T cells expressing CARs targeting B-cell maturation antigen (BCMA) are effective therapy for relapsed or refractory multiple myeloma (MM).21,24 In patients with MM, progression-free survival and overall survival did not differ between patients who received corticosteroids for treatment of CAR T-cell toxicity and those who did not receive corticosteroids.25,27 In addition, the complete response rate was not different for patients who received or did not receive corticosteroids after CAR T-cell therapy for MM.^{25 26} The interplay between corticosteroids and CAR T cells is particularly pertinent to MM because dexamethasone is both a treatment for CAR T-cell toxicities and a standard treatment for MM.14 28,30

We established a murine CRS model to assess the effect of corticosteroids on treatment of MM with anti-BCMA CAR T cells. To our knowledge, this is the first murine study of CRS in the context of MM treatment with CAR T cells. This model used immunodeficient mice that were engrafted with cells of a human MM cell line and treated with a high dose of anti-BCMA CAR T-cells. Because they are a source of IL-6 and other important cytokines, monocyte-lineage cells are critically important for CRS.^{9 12} However, human monocytes only persist in mice when some special strains are engrafted with human hematopoietic stem cells (HSCs).³¹ Therefore, we engrafted mice with THP-1, an acute monocytic leukemia cell line, as a feasible and convenient monocyte substitute for modeling CRS in mice.³² This model has an advantage of human cytokine release from both monocytic cells and CAR T cells; in addition, it includes a human MM cell line. Using this model, we administered anti-BCMA CAR T cells and dexamethasone to mice, and we assessed CRS, antimyeloma activity and in vivo CAR T-cell levels. Dexamethasone alleviated CRS and did not inhibit CAR T-cell antimalignancy activity; in addition, dexamethasone increased CAR T-cell numbers in mice. We also provide data showing blood CAR⁺ cell levels in humans treated with anti-BCMA CAR T cells and corticosteroids.

Methods

Patient samples and mice

Peripheral blood mononuclear cells were obtained from patients with hematological malignancies enrolled in National Cancer Institute (NCI) clinical trials. Human data were obtained from patients enrolled in a clinical trial of a fully human anti-BCMA CAR (ClinicalTrials.gov NCT03602612).²²

Cell lines

Cell lines were obtained from the American Type Culture Collection. RPMI8226, MM.1S, and MM.1R are BCMA⁺ cell lines; MM.1R is resistant to corticosteroid treatment.^{33 34} NGFR-K562 are K562 cells transduced with the gene for low-affinity nerve growth factor.³⁵ THP-1 is an acute monocytic leukemia cell line.³²

CARs, gamma-retroviral transductions, and T-cell culture

Human T cells were used in all experiments. T-cell culture and transduction of T cells with γ-retroviruses encoding CARs were performed as previously described.^{36 37} The anti-BCMA CAR used in this work is designated 11D5-3-CD8BBZ.³⁸ T cells expressing this CAR are called CAR-BCMA. T cells expressing the anti-CD19 CAR Hu19-CD828Z are called CAR-19.³⁹

Flow cytometry

CAR expression was detected by flow cytometry after staining with phycoerythrin-labeled BCMA protein tagged with both fragment crystallizable protein and polyhistidine (BCMA-Fc-His-PE, Creative Biomart).

Cells were stained with the following antibodies from Thermo Fisher Scientific eBioscience: CD3 APC-Cy7 (Clone UCHT1), CD4 BV510 (Clone RPA-T4), CD8 PE-Cy7/eFluor450 (Clone RPA-T8). For memory and exhaustion marker staining, the following antibodies were purchased from Biolegend: CD45RA FITC (Clone HI100), CCR7 PE-Dazzle 594 (Clone G043H7), T-cell immunoglobulin and mucin-domain containing-3 (TIM-3) (Clone F38-2E2), lymphocyte-activation gene (LAG)-3 (Clone 11C3C65), and programmed cell death protein 1 (PD-1) (Clone EH12.2H7). Dead cells were excluded with 7-amino-actinomycin D (BD Biosciences). Flow cytometry was performed with LSRFortessa (BD) or FACSymphony A5 (BD) cytometers. Flow cytometry analysis was performed with FlowJo (Tree Star).

ELISAs and electrochemiluminescence assays

Co-cultures for in vitro ELISA were conducted at 37°C for 18–20 hours as previously described in 96-well plates.⁴⁰ For CAR T cells and cell lines such as THP-1, MM.1S, or RPMI8226, 100 000 cells were added to wells. When primary monocytes were used, 200 000 cells were added because primary monocytes have limited survival, whereas THP-1 cells proliferate. Monocytes were purified with a monocyte purification kit (StemCell Technologies). IFNγ ELISAs were performed on co-culture supernatants with standard methods (Thermo Fisher Scientific). IL-6 and MCP-1 ELISAs were performed on co-culture supernatants with DuoSet ELISA kits (R&D Systems). When noted in figure legends, multiple human cytokines were measured in either culture supernatants or mouse sera with the Meso Scale Discovery V-plex electrochemiluminescence assay platform as previously described.²²

Establishment of a CRS mouse model and dexamethasone treatment

MM.1S and MM.1R cell lines transduced with a gamma-retrovirus encoding the enhanced firefly luciferase gene are referred to as MM.1S-veff-Luc and MM.1R-veff-Luc.⁴¹ NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice were injected intravenously with MM.1S-veff-Luc or MM.1R-veff-Luc cells followed by intravenous injections of THP-1 and CAR T cells as described in figure legends. Following CAR T-cell injection, mice were monitored daily for signs of illness including weight loss, low surface temperature, reduced activity, and changes in fur texture. Mice with >15% weight loss, significantly reduced mobility, or needing gel meals were sacrificed unless they recovered within 36 hours. Temperatures were measured with a Braun infrared thermometer. Mice received an overall CRS grade as described in online supplemental table 1, with 0 being no CRS and 3 being death. Bioluminescent imaging (BLI) of mice was performed on the day of CAR T-cell infusion and every 3 days thereafter as previously described.³⁷ Treatment groups that mice belonged to were blinded to the investigators assessing CRS and tumor burden. Mice received intraperitoneal doses of 5 mg/kg of dexamethasone (dexamethasone sodium phosphate for injection, Somerset Therapeutics) or phosphate-buffered saline (PBS) vehicle control.

Quantitative PCR for CAR⁺ cell quantification

Quantitative PCR (qPCR) to detect human blood CAR⁺ cells was conducted with a method similar to the PCR used for CAR⁺ cell detection in mice and as previously described.²²

For qPCR analysis of CAR⁺ splenocytes in mouse spleens, spleens were mechanically disaggregated and red blood cells were depleted. Splenocytes were counted, and DNA was extracted using a Qiagen DNeasy kit. DNA was amplified in duplicate with a primer and probe set (Thermo) that was specific for the 11D5-3-CD8BBZ CAR that is expressed by CAR-BCMA. Real-time PCR was carried out with a Roche Light Cycler 96 real-time PCR system. Similar to previous work,⁴² we made serial 1:5 dilutions of DNA from the infused CAR T cells of each experiment into untransduced T-cell DNA, and we made standard curves by performing qPCR on this DNA. We determined the percentage of the infused T cells expressing the CAR by flow cytometry as described under the ‘Flow cytometry’ section. To determine the percentage of splenocytes that contained the CAR gene, qPCR results from each mouse were compared with the qPCR results from each experiment’s infused CAR T-cell standard curve. PCR results were normalized to human β-actin. For the actin qPCR, serial dilutions of an actin plasmid (Origene part# SC319328 ACTB NM_001101) were amplified in duplicate with a primer and probe set (Bio-Rad ACTB qHsaCEP0036280) to generate a standard curve. This standard curve was used to determine actin copy numbers in splenocyte DNA samples that had been amplified with the actin primer and probe set. After the percentage of CAR⁺ cells was determined by qPCR, the absolute number of CAR⁺ T cells in each spleen was calculated by multiplying the percentage of CAR⁺ cells by the total number of splenocytes and the fraction of total splenocytes that were CD3⁺. The fraction of splenocytes that were CD3⁺ was determined by flow cytometry.

Statistics

Statistics were performed with GraphPad Prism V.10.1.2. P<0.05 was considered statistically significant. Correction for multiple comparisons was carried out only when noted in figure legends. Details of statistics are provided in figure legends.

Results

Primary monocytes and the monocytic leukemia cell line THP-1 alter cytokine production in CAR T cell plus target cell co-cultures

All of our experiments used human T cells expressing the 11D5-3-CD8BBZ CAR (CAR-BCMA).³⁸ We aimed to establish a murine CRS model incorporating three types of human cells: monocyte-lineage cells, MM cells, and CAR T cells. Obtaining human HSCs is difficult, and monocyte maturation from HSC is blocked at the promonocyte stage in NSG mice, which necessitates use of specialized mouse strains for human monocyte engraftment in mice.³¹ We hypothesized that the monocytic leukemia cell line THP-1 could substitute for primary monocyte-lineage cells in a murine CRS model. Addition of THP-1 or primary monocytes to MM.1S+CAR-BCMA co-cultures caused only modest statistically non-significant increases of IFNγ in co-culture supernatants (figure 1A). MCP-1 was increased in a statistically significant manner by adding either THP-1 or primary monocytes to co-cultures of CAR-BCMA plus MM.1S (figure 1B). IL-6 was increased in a statistically significant manner by adding THP-1 cells to MM.1S+CAR-BCMA co-cultures, and there was a trend toward an increase in IL-6 after addition of primary monocytes to MM.1S+BCMA CAR co-cultures (figure 1C). IL-10 levels were numerically higher after addition of THP-1 or primary monocytes to MM.1S+CAR-BCMA co-cultures, but only primary monocytes caused a statistically significant increase (figure 1D). Other cytokines, including chemokine (C-X-C motif) ligand 9 (CXCL9), IL-17A, GM-CSF, IL-2, and TNF-α did not increase in a statistically significant manner after addition of either THP-1 or primary monocytes to co-cultures of CAR-BCMA plus MM.1S (online supplemental figure 1A-E).

Figure 1. Impact of monocytes and THP-1 cells on in vitro cytokine production. (A) CAR-BCMA cells were co-cultured with MM.1S alone or in combination with either THP-1 or primary monocytes overnight. ELISA was performed on supernatants for (A) IFNγ, (B) MCP-1, and (C) IL-6. Supernatants were assayed for (D) IL-10 by electrochemiluminescence assay. Cells were from four different donors. Statistical analyses were performed using a two-tailed paired ratio t-test with Bonferroni correction. Specifically, p values were all multiplied by three because three comparisons were performed for each cytokine. (E–G) CAR-BCMA were co-cultured overnight with BCMA⁺ MM.1S, BCMA⁺ RPMI-8226, or BCMA⁻ NGFR-K562. Untransduced (UT) T cells and CAR-19 anti-CD19 CAR T cells were included as controls. ELISAs were performed on supernatants for (E) IFNγ, (F) IL-6, and (G) MCP-1. Statistical analyses were performed using a two-tailed paired ratio t-test, without correction for multiple comparisons, and bars represent means+SEM. N=4 different donor PBMCs for all target cells except RPMI8226, which has n=3. Differences were observed in IL-6 release between CAR-BCMA and CAR-BCMA+THP-1 for MM.1S target cells (p=0.0077) and for RPMI8226 target cells (p=0.0303). There were differences in MCP-1 release between CAR-BCMA and CAR-BCMA+THP-1 for MM.1S (p=0.0047) and for RPMI8226 (0.0451). For IL-6, differences were observed between CAR-19 vs CAR-19+THP-1 when target cells were MM.1S (p<0.0001) or RPMI8226 (p<0.0001). For MCP-1, differences were observed between CAR-19 vs CAR-19+THP-1 when target cells were MM.1S (p=0.0047) or RPMI8226 (p=0.0183). For the comparison of IL-6 release between UT and UT+THP-1, p=0.0637 when target cells were MM.1S, and p=0.0296 when target cells were RPMI8226. For MCP-1, differences were observed between UT versus UT+THP-1 when target cells were MM.1S (p<0.0001) or RPMI8226 (p=0.0332). *P<0.05; **p<0.01. BCMA, B-cell maturation antigen; CAR, chimeric antigen receptor; IFN, interferon; IL, interleukin; MCP-1, monocyte chemoattractant protein-1; MM, multiple myeloma; PBMC, peripheral blood mononuclear cell; THP-1, acute monocytic leukemia cell line; UT, untransduced T cells.

IFNγ was released in an antigen-specific manner in co-cultures of CAR-BCMA plus BCMA⁺ target cells MM.1S or RPMI8226, but THP-1 cells did not influence IFNγ levels (figure 1E). Addition of THP-1 to co-cultures of T cells plus target cells caused statistically significant increases in IL-6 and MCP-1 (figure 1F, G). Levels of IL-6 and MCP-1 in co-culture supernatants were highest when both THP-1 and CAR-BCMA were included in co-cultures, indicating that recognition of MM.1S by CAR-BCMA led to increased IL-6 and MCP-1 levels (figure 1F, G). In co-cultures containing MM.1S and THP-1, the median IL-6 level was 907 pg/mL (453–1012 pg/mL) with CAR-BCMA vs 168 pg/mL (94–210 pg/mL) with CAR-19 (p=0.0114, ratio paired t-test), and the median MCP-1 level was 53 161 pg/mL (7532–87 305 pg/mL) with CAR-BCMA vs 7849 pg/mL (3679–9849 pg/mL) with CAR-19 (p=0.0173, ratio paired t-test).

Dexamethasone reduced cytokine levels in vitro

Dexamethasone is a corticosteroid used clinically to manage CAR T-cell toxicities.^{3 4} To evaluate the effect of dexamethasone on cytokine production in vitro, we set up co-cultures containing CAR-BCMA and MM.1S with or without THP-1. Dexamethasone caused a statistically significant reduction of IL-6 and MCP-1 in supernatants of co-cultures of CAR-BCMA, MM.1S, and THP-1 (figure 2A). Adding dexamethasone to co-cultures of CAR-BCMA, MM.1S, and THP-1 did not decrease IFNγ levels in a statistically significant manner.

Figure 2. Dexamethasone decreased in vitro cytokine release and signs of CRS in mice. (A) On day 9 of T-cell culture, untransduced control T cells or CAR-BCMA were co-cultured overnight with MM.1S cells with or without THP-1. Co-cultures were supplemented with either 0.5 μM dexa or veh control. ELISAs were performed on supernatants for IFNγ, IL-6, and MCP-1. Bars represent mean+SEM. Statistical analyses were performed using a two-tailed paired ratio t-test; n=4 different donor PBMCs, except MCP-1 untransduced+MM.1S+THP-1+veh had three donor PBMCs. P values for CAR-BCMA+THP-1+MM.1S co-cultures with dexa versus veh are shown on the graphs. There were not statistically significant differences in cytokine release with dexa versus veh for cultures that did not contain THP-1. Some MCP-1 bars are not visible due to very low levels. (B) Schema of CRS model and dexa treatment protocol. Female NSG mice received 10×10⁶ MM.1S-veff-Luc cells i.v. Four days later, mice received 10×10⁶ THP-1 cells i.v. Eighteen days after the initial MM.1S-veff-Luc infusion, mice received 10×10⁶ CAR-BCMA i.v. or were left untreated. The day of CAR-BCMA infusion was defined as day 0. Either veh or 5 mg/kg dexa was administered intraperitoneally on days ‘D’ 1, 3, and 5 after CAR-BCMA infusion. Tumor burden was assessed every 3 days by BLI. From the day of T-cell infusion, mice were monitored for CRS indicators including weight, temperature, fur texture, and activity. (C) Median±IQR of overall CRS grades of dexa and veh groups are shown on the indicated days after CAR-BCMA infusion. Overall CRS grades were higher for veh mice versus dexa mice on days 5, 6, 8, and 9. P<0.0008 by two-tailed Mann-Whitney U test plus Bonferroni correction at the time points indicated by asterisks. Statistical analysis was performed on data from all eight CRS assessment time points after CAR-BCMA infusion. Bonferroni correction was performed by multiplying the Mann-Whitney p values by 8. Results are from four separate experiments, each using cells from a different donor. The total number of mice in each of the veh and dexa groups was 15. (D) Overall CRS grades of the same mice as in (C) were compared on day 8 after CAR T-cell infusion; this time point was chosen because it was the last day when both groups had evidence of CRS. Black bars show the medians. P<0.0008 by Mann-Whitney U test after Bonferroni correction. BCMA, B-cell maturation antigen; BLI, bioluminescence imaging; CAR, chimeric antigen receptor; CRS, cytokine-release syndrome; dexa, dexamethasone; IFN, interferon; IL, interleukin; i.v., intravenously; MCP-1, monocyte chemoattractant protein-1; MM, multiple myeloma; NSG, NOD-scid common γ-chain-deficient mice; PBMC, peripheral blood mononuclear cell; THP-1, acute monocytic leukemia cell line; veh, vehicle.

Establishment of a murine CRS model

The knowledge that THP-1 cells lead to increased release of IL-6 and MCP-1 in vitro led us to include intravenous injections of THP-1 in a mouse model of CRS. The model also included intravenous injections of MM.1S-veff-Luc cells and CAR-BCMA (figure 2B). Use of the MM.1S-veff-Luc MM cell line allowed monitoring of malignancy burdens by BLI.⁴¹ Doses of MM.1S-veff-Luc cells and CAR-BCMA in this model were higher than doses in our prior studies.^{37 43} Serum levels of IL-6 and MCP-1 were consistently higher in mice that received THP-1 plus MM.1S-veff-Luc and CAR-BCMA compared with mice that received only MM.1S-veff-Luc and CAR-BCMA, although differences only reached p<0.05 for MCP-1 three to four days after CAR-BCMA infusion (online supplemental figures 1F-H, 2A). When compared with mice that received only MM.1S-veff-Luc and CAR-BCMA, mice that received THP-1 plus MM.1S-veff-Luc and CAR-BCMA had larger temporary weight decreases during the time of CRS (online supplemental figure 2B). This indicated more severe CRS when THP-1 was included. Of note, 10 of 10 mice injected with 10×10⁶ THP-1 cells alone survived until 40 days after injection without apparent toxicity, so THP-1 cells did not cause early deaths or toxicity during CRS assessment.

We graded murine CRS severity on a 0–2 grading scale based on four parameters: per cent weight loss, surface temperature, activity, and fur texture, as described in online supplemental table 1. In this system, mouse death was grade 3 CRS.

To assess for the antigen specificity of our model, we conducted an experiment in which mice were engrafted with both MM.1S-veff-Luc and THP-1 cells and then treated with either CAR-BCMA or CAR-19 (irrelevant anti-CD19 CAR); another group received no T cells (online supplemental figure 3). Mice in the CAR-BCMA group showed signs of CRS consisting of low body temperature, weight loss, and ruffled fur 1 and 2 days after CAR-T cell infusion; in contrast, no signs of CRS were apparent in the CAR-19 and the No T cells groups on these days (online supplemental figure 3A). Three days after CAR T-cell infusion, the CAR-BCMA group continued to have a statistically significantly higher median overall CRS grade compared with the CAR-19 and No T cells groups. Four days after CAR T-cell infusion, four of five mice in the CAR-BCMA group died. Because of the known activity of CAR-BCMA against MM.1S and signs of CRS that the mice were exhibiting, these deaths were likely caused by CRS. Throughout the experiment, signs of illness exhibited by the CAR-19 and No T cells groups were only weight loss and hind-limb paresis, which were likely due to rapid progression of MM.1S in these mice. The CAR-19 and No T cells groups had statistically longer survival compared with mice in the CAR-BCMA group (online supplemental figure 3B). In summary, signs of severe CRS required inclusion of CAR-BCMA cells.

Tumor lysis syndrome, which is caused by destruction of malignant cells, can cause signs of illness that partially overlap with signs of CRS. An elevated level of blood uric acid is an important marker of tumor lysis syndrome.⁴⁴ The signs of illness exhibited by mice in our CRS model were unlikely to be from tumor lysis syndrome because blood uric acid was lower in mice that received injections of MM.1S-veff-Luc, THP-1, and CAR-BCMA versus unmanipulated mice that received no cell infusions (online supplemental figure 4A).

Treatment of murine CRS with dexamethasone

We used the CRS model of NSG mice injected with MM.1S-veff-Luc, THP-1, and CAR-BCMA (figure 2B) to assess CRS treatment with dexamethasone. Mice were treated with either 5 mg/kg dexamethasone or the PBS vehicle control on days 1, 3, and 5 after CAR-BCMA infusion. When overall CRS grades of mice receiving vehicle control or dexamethasone were compared, mice treated with dexamethasone had lower median overall CRS grades 5, 6, 8, and 9 days after CAR-BCMA infusion, meaning dexamethasone reduced CRS severity (figure 2C, D). In these experiments, mice started showing signs of illness approximately 3 days after CAR-BCMA infusion, and illness became severe in mice not treated with dexamethasone on days 5–6 after CAR-BCMA infusion (figure 2C). In figure 2C, dead mice counted as an overall CRS grade of 3 at every time point after death. Mice receiving dexamethasone maintained more of their baseline body weight when compared with mice receiving vehicle control (figure 3A, online supplemental figure 4B). Overall, these results show that dexamethasone significantly alleviated clinical signs of illness in this CRS model.

Figure 3. Dexamethasone decreased weight loss and serum cytokine levels, which are manifestations of CRS. For all of figure 3, MM.1S-veff-Luc cells and THP-1 cells were infused, and CAR-BCMA T cells were administered as described in figure 2B. (A) Baseline weights of mice were recorded immediately before CAR-BCMA infusion on day 0. Mean baseline weights for each group were considered 100%. Weights were monitored for 11 days after infusion. Weights are shown as mean percentages of baseline weights±SEM for each group. Mice were from four separate experiments, and the mice were the same mice as in figure 2C–D. For each group, the following numbers of mice were analyzed at each time point: CAR-BCMA+dexa, 14–15; CAR-BCMA+veh, 9–15; No T cells+dexa, 10–11; No T cells+veh, 11. Numbers have ranges because mice died in some groups during the experiments. The No T cells+dexa and No T cells+veh curves end at day 6 because mice in these groups were euthanized due to high MM.1S-veff-Luc burdens. *P=0.0089 for the comparison of the entire curves of CAR-BCMA+dexa versus CAR-BCMA+veh by mixed-effects analysis. (B) Cytokines associated with CRS were measured in sera of mice. Serum samples were drawn 6 days after CAR-BCMA infusion. Groups receiving veh control or dexa are indicated on the x-axis of each graph. Cytokines assayed are on the y-axis of each graph. All proteins were measured by electrochemiluminescence assay except IFNγ and GM-CSF, which were measured by ELISA. P values are shown on the plots. Statistical analyses were performed using a two-tailed paired ratio t-test. Three experiments were performed for each cytokine except GM-CSF, which had four experiments. Each experiment used CAR-BCMA from a different donor. BCMA, B-cell maturation antigen; CAR, chimeric antigen receptor; CRS, cytokine-release syndrome; dexa, dexamethasone; GM-CSF, granulocyte-macrophage colony-stimulating factor; IFN, interferon; IL, interleukin; MCP-1, monocyte chemoattractant protein-1; MM, multiple myeloma; THP-1, acute monocytic leukemia cell line; TNF-α, tumor necrosis factor-α; veh, vehicle.

Dexamethasone and blood cytokine levels

We used the CRS model described in figure 2B to assess levels of cytokines in sera of mice that were treated with either dexamethasone or vehicle control. The cytokines assessed have been previously found to be elevated in the blood of humans with CRS after CAR T-cell infusions.^{1 4 7} We found that serum levels of IFNγ, GM-CSF, TNF-α, MCP-1, IL-6, macrophage inflammatory protein-1α (MIP-1α), IL1β, IL-2, IL-10, and IL-8 were all detectable in sera of mice 6 days after CAR-BCMA infusion (figure 3B). There was a consistent trend toward decreased levels of most cytokines when mice were treated with dexamethasone, but the decreased levels only reached statistical significance for IFNγ.

Dexamethasone did not impair MM.1S clearance

To assess whether dexamethasone impacts CAR T-cell antimalignancy efficacy, we used the MM.1S-veff-Luc+THP-1+CAR-BCMA model (figure 2B) to evaluate clearance of MM.1S-veff-Luc cells. Mice treated with CAR-BCMA, either with or without dexamethasone, effectively cleared MM.1S-veff-Luc cells (figure 4A). Figure 4A shows results of one representative experiment of four experiments. BLI values of all mice from all four separate experiments are shown in figure 4B. When we assessed all mice by BLI on day 6 after CAR-BCMA infusion, mice treated with the combination of CAR-BCMA and dexamethasone had lower malignancy burdens compared with mice treated with CAR-BCMA and the vehicle control (figure 4C). Day 6 after CAR-BCMA infusion was chosen for analysis because this was the last day when mice from all four experiments still had a measurable malignancy burden. In addition, the percentage of mice surviving was higher in a statistically significant manner in mice receiving dexamethasone versus vehicle control (figure 4D). This higher survival occurred because there was a lower death rate from CRS in mice receiving dexamethasone. The same mice were the source of CRS data in figures2C, D 3A as well as antimyeloma activity data in figure 4A–D.

Figure 4. Effect of dexamethasone on MM elimination and blood CAR T-cell numbers. For all of figure 4, MM.1S-veff-Luc cells, THP-1 cells, CAR-BCMA, and dexa were infused as described in figure 2B. The same CAR-BCMA+veh control and CAR-BCMA+dexa mice were the sources of data for figures 2C–D, 3A, and 4A–D. Veh and dexa groups were compared. (A) Tumor burden was assessed by BLI every 3 days. BLI images of one representative experiment are shown. Four similar experiments were conducted. (B) The radiance of each individual mouse in all treatment groups and all four experiments of this type is represented by individual lines. Symbols in the legend indicate the treatments received by the different groups. N=11 for both No T cells+veh and No T cells+dexa. N=15 for both CAR-BCMA+veh and CAR-BCMA+dexa. (C) Tumor burdens were compared between CAR-BCMA+veh and CAR-BCMA+dexa groups on day 6 after CAR-BCMA infusion. These are data from day 6 of the same mice as in 4B. There was a statistically significant difference by two-tailed unpaired t-test. Each group had a total of 15 mice from four experiments with T cells from different donors. P value is on the graph; horizontal bars are at the medians. (D) Kaplan-Meier survival plot of the same mice as in 4B-C. There was a statistically significant difference in survival between the CAR-BCMA+veh and CAR-BCMA+dexa groups (p=0.0347, log-rank test). (E–J) Mice were sacrificed 13 days after CAR-BCMA infusion, and splenocytes were isolated. Live splenocytes were counted and stained for CD3, CD4, CD8, and CAR. Absolute cell counts were calculated for each T cell subset, and comparisons were made between the veh control and dexa-treated groups: (E) total CD3⁺ T cells, (F) total CD3⁺CD4⁺ T cells, (G) total CD3⁺CD8⁺ T cells, (H) CD3⁺ CAR⁺ T cells, (I) CD3⁺CD4⁺CAR⁺ T cells, and (J) CD3⁺CD8⁺CAR⁺ T cells. Each symbol represents the result from an individual mouse. Statistical analyses were performed using a two-tailed Mann-Whitney U test. P values are on graphs; horizontal bars are at the medians. N=8 for the veh group, and n=10 for the dexa group. BCMA, B-cell maturation antigen; BLI, bioluminescence imaging; CAR, chimeric antigen receptor; dexa, dexamethasone; MM, multiple myeloma; veh, vehicle.

Dexamethasone was associated with higher levels of total T cells and CAR-BCMA

We assessed splenic T cells in mice prepared as described in figure 2B. We sacrificed mice 13 days after CAR-BCMA infusion and evaluated splenic T cells by flow cytometry. Dexamethasone-treated mice had higher absolute numbers of total CD3⁺, total CD3⁺CD4⁺, and total CD3⁺CD8⁺ T cells when compared with vehicle control-treated mice (figure 4E–G). Similarly, dexamethasone-treated mice had higher absolute numbers of CD3⁺CAR⁺ T cells (figure 4H) and CD3⁺CD8⁺CAR⁺ T cells (figure 4J). There was a statistically non-significant trend toward higher absolute numbers of CD3⁺CD4⁺CAR⁺ T cells in dexamethasone-treated mice versus vehicle control-treated mice (figure 4I). In addition, we found that CD3⁺, CD3⁺CD4⁺, and CD3⁺CD8⁺ CAR-negative T-cell numbers were higher in a statistically significant manner in mice receiving dexamethasone versus mice receiving the vehicle control (online supplemental figure 5). In a different experiment, we prepared mice as described in figure 2B, and we sacrificed these mice 7 days after CAR-BCMA infusion. There were statistically non-significant trends toward higher absolute numbers of splenic CD3⁺CAR⁺ cells and CD3⁺CD4⁺CAR⁺ cells in mice receiving dexamethasone versus mice receiving the vehicle control (online supplemental figure 6). In addition, we performed flow cytometry to assess markers of memory and exhaustion 7 days after CAR-BCMA infusion (online supplemental figures 7A-E, 8A-E). We found a lower median percentage of CD3⁺CD8⁺CAR⁺ cells with a phenotype of terminally differentiated effector memory T cells re-expressing CD45RA (TEMRA) in mice receiving dexamethasone versus mice receiving vehicle control (online supplemental figure 8B). We also found lower expression of the exhaustion marker LAG-3 on CD3⁺CD8⁺CAR⁺ cells among mice receiving dexamethasone versus mice receiving vehicle control (online supplemental figure 8C).

Treatment of a dexamethasone-resistant MM cell line with CAR-BCMA plus dexamethasone

Because corticosteroids such as dexamethasone have activity against MM, some of the antimalignancy activity in our experiments using MM.1S-veff-Luc could have been due to dexamethasone having a direct effect on MM.1S-veff-Luc cells. To address this, we conducted experiments with the dexamethasone-resistant cell line MM.1R-veff-Luc.³⁴ In these experiments, mice were injected with bioluminescent MM.1R-veff-Luc cells followed by THP-1 cells. Mice were then infused with CAR-BCMA 25 days after MM.1R-veff-Luc injection (figure 5A). In this model, CAR-BCMA cells eliminated MM.1R-veff-Luc cells at a faster rate when dexamethasone was administered compared with when a vehicle-only negative control was administered (figure 5B–D). These results were similar to the results of CAR-BCMA plus dexamethasone treatment of dexamethasone-sensitive MM.1S-veff-Luc (figure 4A–C). Importantly, dexamethasone did not impair long-term elimination of MM.1R-veff-Luc cells (figure 5E). For unknown reasons, CRS was much less severe with MM.1R-veff-Luc compared with MM.1S-veff-Luc, so the MM.1R-veff-Luc model was not suitable for CRS comparison between dexamethasone and vehicle control groups. Online supplemental figure 9 summarizes overall CRS grades for the MM.1R-veff-Luc experiments 5–6 days after CAR-BCMA infusion when CRS was at peak severity in these experiments.

Figure 5. Evaluation of the antimalignancy activity of CAR-BCMA plus dexamethasone against the dexamethasone-resistant MM.1R-veff-Luc cell line. (A) For all experiments in figure 5, female NSG mice were injected with 10×10⁶ MM.1R-veff-Luc tumor cells i.v. Seven days later, on day −18, mice received 10×10⁶ THP-1 cells i.v. Twenty-five days after the MM.1R-veff-Luc infusion, mice received 10×10⁶ CAR-BCMA i.v. In addition, control groups received no T cells. Either veh or dexa was administered i.p. on days ‘D’ 1, 3, and 5 after CAR-BCMA infusion. MM.1R-veff-Luc burden was assessed every 3 days by BLI. (B) BLI images show mice in different treatment groups. This is a representative example of two experiments with similar results. Mice were sacrificed on day 13 for T-cell studies on splenocytes. (C) The radiance of each individual mouse in all treatment groups and both experiments is represented by individual lines. Symbols in the legend indicate the treatments received by the different groups. N=10 for all groups. (D) Malignancy burdens of the CAR-BCMA+veh and CAR-BCMA+dexa groups were compared on day 6 after CAR-BCMA infusion, which was the last day that MM.1R-veff-Luc cells were detectable. Each symbol in the graph represents the BLI measurement from an individual mouse; horizontal bars are at the medians. The CAR-BCMA+veh and CAR-BCMA+dexa groups each had 10 mice. Statistical analyses were performed using an unpaired, two-tailed t-test; p value is on the graph. Combined results from two independent experiments are shown. The same mice are reported in C and D, and images in B are from one of two experiments reported in C and D. (E) In a separate experiment from the mice reported in 5B–D, mice treated as described in 5A were followed for long-term tumor control by BLI. All mice treated in this experiment are shown. BCMA, B-cell maturation antigen; BLI, bioluminescence imaging; CAR, chimeric antigen receptor; dexa, dexamethasone; i.v., intravenously; MM, multiple myeloma; NSG, NOD-scid common γ-chain-deficient mice; THP-1, acute monocytic leukemia cell line; veh, vehicle.

Dexamethasone increased CAR T-cell numbers in a dexamethasone-resistant MM model

To assess CAR T-cell levels in mice bearing MM.1R-veff-Luc cells, mice were prepared as described in figure 5A. One group was treated with dexamethasone, and another group was treated with vehicle control. Mice were sacrificed 13 days after CAR-BCMA infusion, and splenocytes were isolated and analyzed by flow cytometry. There was a nonstatistically significant trend toward higher total human CD3⁺ cell numbers with dexamethasone injections versus vehicle control injections (figure 6A). As in the MM.1S-veff-Luc mouse studies, there were higher numbers of total human CD3⁺CD4⁺ T cells and CD3⁺CD8⁺ T cells in mice receiving dexamethasone versus mice receiving vehicle control (figure 6B, C). The numbers of splenic CD3⁺CAR⁺ T cells, CD3⁺CD4⁺CAR⁺ T cells, and CD3⁺CD8⁺CAR⁺ T cells were statistically significantly higher in mice receiving dexamethasone compared with mice receiving vehicle control (figure 6D–F). There was not a statistically significant difference in CD4:CD8 ratios of CD3⁺CAR⁺ T cells when the dexamethasone and vehicle control groups were compared (figure 6G). To validate the results from flow cytometry, a CAR-BCMA-specific qPCR assay was used to quantify CAR⁺ splenocytes. In agreement with results from flow cytometry, qPCR confirmed that the number of CAR⁺ T cells was higher in the dexamethasone group versus the vehicle control group (figure 6H).

Figure 6. CAR T-cell levels were higher with dexamethasone treatment in mice bearing MM.1R-veff-Luc. Mice from experiments reported in figure 5C,D were sacrificed 13 days after CAR-BCMA infusion, and splenocytes were isolated. (A-F) Total live splenocytes were stained for CD3, CD4, CD8, and CAR to quantify T-cell subsets and CAR expression. Plots show the absolute cell counts calculated for each T-cell subset. Veh-treated and dexa-treated groups were compared: (A) total CD3⁺ T cells, (B) total CD3⁺CD4⁺ T cells, (C) total CD3⁺CD8⁺ T cells, (D) CD3⁺CAR⁺ T cells, (E) CD3⁺CD4⁺CAR⁺ T cells, and (F) CD3⁺CD8⁺CAR⁺ T cells. (G) CD4:CD8 ratios of CD3⁺CAR⁺ T cells from splenocytes of the veh and dexa groups are shown. (H) DNA was extracted from the total splenocytes of each mouse. qPCR was performed by using a primer-probe set specific for CAR-BCMA, and the absolute numbers of CAR⁺ T cells were calculated. For all of figure 6, each symbol on the graphs represents the result from an individual mouse, and n=10 mice per group. Horizontal bars represent medians. Statistics were Mann-Whitney U tests. P values are shown on the graphs. BCMA, B-cell maturation antigen; CAR, chimeric antigen receptor; dexa, dexamethasone; MM, multiple myeloma; qPCR, quantitative PCR; veh, vehicle.

High levels of blood CAR⁺ cells were achieved in human patients with MM receiving corticosteroid treatment

We assessed blood CAR⁺ cell levels during and after corticosteroid administration in the four patients with MM who received corticosteroids on a recent clinical trial of an anti-BCMA CAR.²² Blood CAR⁺ cells were quantified with a qPCR assay that was specific for the CAR used in the clinical trial. The four patients who received corticosteroids all had peak blood CAR⁺ cell levels of >100 CAR⁺ cells/μL, and in all four patients, CAR⁺ cell levels continued to increase after corticosteroid therapy was initiated (figure 7A–D). In most clinical trials, CAR⁺ cell levels peak between 7 and 15 days after CAR T-cell infusion,^{6 22 42 45} and the results of the four patients that we analyzed followed this pattern (figure 7A–D). Figure 7E compares blood CAR⁺ cell levels of patient 12 with the other two patients who received the same dose of 3×10⁶ CAR⁺ T cells/kg on the same clinical trial. Patient 12 received corticosteroids (figure 7A), but patients 10 and 11 did not. Blood CAR⁺ cell levels of patients 13, 20, and 21 were compared with blood CAR⁺ cell levels of other patients who received the same dose of 6×10⁶ CAR⁺ T cells/kg on the same clinical trial. Patients 13, 20, and 21 received corticosteroids (figure 7B–D). The other patients shown in figure 7F did not receive corticosteroids. Clinical outcomes and details of corticosteroid administration are provided in online supplemental table 2.

Figure 7. High levels of blood CAR⁺ cells could be achieved in humans receiving corticosteroids. (A–D) Blood CAR⁺ cells were quantified by qPCR in the four patients who received corticosteroids to treat adverse events on an anti-BCMA CAR T-cell clinical trial. All data in this figure are from the same clinical trial. Line graphs show CAR⁺ cell counts per µL of blood following CAR T-cell infusion. Red arrows indicate the days when each patient received corticosteroids. (A) Patient 12 received corticosteroids on days 2 and 3 after CAR T-cell infusion. (B) Patient 13 received corticosteroids on days 5–11 after CAR T-cell infusion. Note that patient 13 also received radiation therapy during the same time that he received corticosteroids. (C) Patient 20 received corticosteroids on day 1 after CAR T-cell infusion and (D) patient 21 received corticosteroids on days 3 and 4 after CAR T-cell infusion. (E) The graph shows blood CAR⁺ cell counts of all patients who received 3×10⁶ CAR T cells/kg, which was the same dose as patient 12. (F) The graph shows blood CAR⁺ cell counts of all patients who received 6×10⁶ CAR T cells/kg, which was the same dose as patients 13, 20, and 21. In E–F, patients who received corticosteroids are in red in the legends of the graphs. BCMA, B-cell maturation antigen; CAR, chimeric antigen receptor; qPCR, quantitative PCR.

Discussion

CAR T-cell therapy has improved treatment of many hematological malignancies including MM.^{6 21 45} CRS is a common toxicity of CAR T-cell therapy.^{1 4 21} CRS can be treated with drugs including tocilizumab and corticosteroids.^{3 4} Corticosteroids are also critical for treating CAR-associated neurological toxicity.^{3 4} Therefore, learning more about the interaction between CAR T cells and corticosteroids is an important area of study.

Addition of human dendritic cells or macrophages to in vitro co-cultures of anti-CD19 CAR T cells and an acute lymphoblastic leukemia cell line greatly increased levels of GM-CSF, IL-8, and IL-6 in co-culture supernatants.⁹ Murine models of CRS have previously been established.^{11 12} One previously reported model used SCID-beige mice that were engrafted with a human lymphoma cell line and treated with human CAR T cells.¹¹ In this model, the myeloid cells were of murine origin while the CAR T cells were of human origin.¹¹ This could be a limitation because of the known lack of cross-reactivity between some murine and human cytokines.⁴⁶ Another model used myeloid growth-factor-transgenic mice engrafted with human cord blood hematopoietic stem and progenitor cells.¹² These mice were then engrafted with human leukemia followed by infusion of CAR T cells derived from different humanized mice.¹² While this model is quite elegant, it is also complex and expensive for most investigators. A prior report described treating murine melanoma with syngeneic T cells that were transgenic for a T-cell receptor that recognized a melanoma-associated antigen.⁴⁷ This report found that dexamethasone administration did not reduce T-cell antimelanoma activity and did not decrease levels of activated antimelanoma T cells in mice.⁴⁷

We established a model that used the common NSG mouse strain. The malignancy burden of the model was a bioluminescent human MM cell line, either MM.1S-veff-Luc or MM.1R-veff-Luc, and we treated the mice with human T cells expressing an anti-BCMA CAR. We also engrafted mice with the human THP-1 monocytic leukemia cell line as a substitute for human monocytes and macrophages because both primary monocytes and THP-1 cells caused increases in IL-6 and MCP-1 when added to in vitro co-cultures of CAR-BCMA and MM.1S cells (figure 1B, C). Blood IL-6 and MCP-1 are known to be elevated in humans during CAR toxicity.⁴ Mice in this model exhibited clinical signs of inflammation and elevated levels of serum cytokines that have been previously reported to be elevated in the blood of humans with CRS.^{4 7} Our model incorporates human cells for the three cell types known to be critical for CRS after CAR T-cell infusion: monocyte-lineage cells, CAR T cells, and malignant cells expressing the target of the CAR. In addition, our model is feasible for many investigators to use.

We found that administering dexamethasone after CAR-BCMA infusions led to amelioration of CRS and a faster eradication of the MM.1S-veff-Luc dexamethasone-sensitive MM cell line from mice. Similar results were obtained when mice bearing MM.1R-veff-Luc,³⁴ the dexamethasone-resistant variant of MM.1S-veff-Luc, were treated with CAR-BCMA plus dexamethasone. This shows that continued antimyeloma activity during CAR-BCMA plus dexamethasone administration was not attributable to direct antimyeloma activity of dexamethasone. One possible explanation why dexamethasone did not inhibit antimyeloma activity of CAR-BCMA is that dexamethasone treatment increased the number of CAR-BCMA in mice.

One possible reason for the association between administration of dexamethasone and higher CAR T-cell levels is a previously reported reduction of activation-induced T-cell death by dexamethasone.⁴⁸ Also, dexamethasone has been shown to upregulate IL-7 receptor-α on CAR T cells,⁴⁹ and administering the combination of dexamethasone and exogenous IL-7 increased blood CAR T-cell levels in mice; in addition, adding dexamethasone and exogenous IL-7 to CAR T infusions modestly improved treatment of ovarian carcinoma in mice.⁵⁰ When we assessed splenic T cells 7 days after CAR-BCMA infusion, dexamethasone administration was associated with lower percentages of splenic CD3⁺CD8⁺CAR⁺ cells that expressed the exhaustion marker LAG-3 or that possessed the terminally differentiated TEMRA phenotype.⁵¹

Some clinical reports showed increased blood CAR T-cell levels in patients receiving corticosteroids.^{19 26} We assessed blood CAR⁺ cell levels of four patients who received corticosteroids on a recent anti-BCMA CAR clinical trial.²² We found, in this small number of patients, that CAR⁺ cell levels continued to increase after administration of corticosteroids (figure 7). In addition, peak blood CAR T-cell levels and CAR T-cell persistence were similar in patients who did or did not receive corticosteroids. A high peak level of blood CAR T-cells is associated with CRS and CAR-associated neurological toxicity, so patients with high peak blood CAR T-cell levels might be more likely to receive corticosteroids.⁴

Our findings suggest that further clinical studies are needed to determine clinical corticosteroid regimens that do not impair, or potentially even enhance, CAR T-cell therapy. Corticosteroids are effective therapy for CRS and CAR-associated neurologic toxicity. CRS and neurologic toxicity can be dose-limiting for CAR T cells. Perhaps corticosteroid administration could allow patients to tolerate higher doses of CAR T cells. Antimalignancy activity in mice generally correlates with CAR T-cell doses,^{36 40 43} and higher CAR T-cell doses have been more effective in some human clinical trials.^{23 24 42} Increasing CAR T-cell doses might be one way to improve the efficacy of CAR T-cell therapies. However, dose, duration, and timing of corticosteroid administration after CAR T-cell infusions must be chosen carefully because high cumulative corticosteroid doses might impair antimalignancy outcomes.¹⁸

The findings and conclusions presented in this paper are those of the authors and do not necessarily reflect the views of the NIH or the US Department of Health and Human Services. The contributions of the NIH authors are considered Works of the United States Government.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.