CD19/22 CAR T cells in children and young adults with B-ALL: phase 1 results and development of a novel bicistronic CAR

Abstract

Remission durability following single-antigen targeted chimeric antigen receptor (CAR) T-cells is limited by antigen modulation, which may be overcome with combinatorial targeting. Building upon our experiences targeting CD19 and CD22 in B-cell acute lymphoblastic leukemia (B-ALL), we report on our phase 1 dose-escalation study of a novel murine stem cell virus (MSCV)-CD19/CD22-4-1BB bivalent CAR T-cell (CD19.22.BBζ) for children and young adults (CAYA) with B-cell malignancies. Primary objectives included toxicity and dose finding. Secondary objectives included response rates and relapse-free survival (RFS). Biologic correlatives included laboratory investigations, CAR T-cell expansion and cytokine profiling. Twenty patients, ages 5.4 to 34.6 years, with B-ALL received CD19.22.BBζ. The complete response (CR) rate was 60% (12 of 20) in the full cohort and 71.4% (10 of 14) in CAR-naïve patients. Ten (50%) developed cytokine release syndrome (CRS), with 3 (15%) having ≥ grade 3 CRS and only 1 experiencing neurotoxicity (grade 3). The 6- and 12-month RFS in those achieving CR was 80.8% (95% confidence interval [CI]: 42.4%-94.9%) and 57.7% (95% CI: 22.1%-81.9%), respectively. Limited CAR T-cell expansion and persistence of MSCV-CD19.22.BBζ compared with EF1α-CD22.BBζ prompted laboratory investigations comparing EF1α vs MSCV promoters, which did not reveal major differences. Limited CD22 targeting with CD19.22.BBζ, as evaluated by ex vivo cytokine secretion and leukemia eradication in humanized mice, led to development of a novel bicistronic CD19.28ζ/CD22.BBζ construct with enhanced cytokine production against CD22. With demonstrated safety and efficacy of CD19.22.BBζ in a heavily pretreated CAYA B-ALL cohort, further optimization of combinatorial antigen targeting serves to overcome identified limitations (www.clinicaltrials.gov #NCT03448393).

Graphical abstract

Figure 1.

Response, outcomes, CAR expansion, and cytokine levels after CD19.22.BBζ CAR T-cell infusion. (A) Waterfall plot of the best response after CD19.22.BBζ CAR T-cell infusion. Participants were stratified by DL and CRS grade. Bone marrow clearance demonstrates those that were MRD negative by flow cytometry. (B) RFS was calculated from the date of CAR infusion until the date of relapse or last follow-up among those who went into complete remission and was not censored for HSCT. The 6- and 12-month RFS was 80.8% (95% CI: 42.4%-94.9%) and 57.7% (95% CI: 22.1%-81.9%), respectively. (C) OS was calculated from the date of CAR infusion until the date of death or last date of follow-up. The 6- and 12-month OS was 100% and 63.0% (95% CI: 35.4%-81.4%) respectively. (D) Duration of continuous remission among patients who achieved a CR. Shown are duration of remission, time to transplant if applicable, and time of relapse denoted as either antigen negative or antigen positive. (E) Peak CAR T-cell expansion in the PB assessed as absolute CAR T cells based on the percent of absolute lymphocyte count that was CAR T cell positive and stratified by CR vs no CR, and patients who were CAR naïve and CAR pretreated are denoted. (F) Peak percentage of CAR T cells in the bone marrow at day 28 assessment stratified by CR vs no CR, and patients who were CAR naïve and CAR pretreated are denoted. (G-J) Comparison of peak cytokine values of IFNγ, IL18, IL2, and GM-CSF between CR and no CR, with CAR naïve denoted with a red circle and CAR pretreated denoted with a blue triangle.

Figure 2.

Comparison of CAR T-cell expansion, persistence, inflammatory markers, and cytokine values across 3 Pediatric Oncology Branch clinical trials. (A-B) Peak CAR T-cell expansion in the PB displayed as absolute CAR T cells and peak percentage of CAR T cells in the bone marrow, respectively, evaluated in responders across the 3 trials. (C) Persistence of CAR T cells in the PB as measured by flow cytometry. (D-E) Comparison of peak CRP and ferritin values across trials. (F-N) Comparison of peak serum cytokine values of tumor necrosis factor α, IL2, IL1B, IL4, IL12p70, IFNγ, IL6, IL8, and GM-CSF across all 3 trials. (O-S) Comparison of peak serum cytokine values of IL10, IL18, macrophage inflammatory protein-1 (MIP1)α, IL15, and IL13 between CD19/22 and CD22 CAR patients. *P > .01 to < .05; **P > .001 to < .01; ***P > .0001 to < .001; ****P < .0001.

Figure 3.

Efficacy of MSCV-CD19.22.BBζ CAR and EF1α-CD19.22.BBζ CAR constructs. (A) Cytokine production by EF1α-CD19.22.BBζ and MSCV-CD19.22.BBζ CAR T cells was evaluated on coculture with the indicated CD19⁺CD22⁺, CD19⁺CD22⁻, CD19⁻CD22⁺, and CD19⁻CD22⁻ NALM6 leukemia lines. Cocultures were performed at a 1:1 effector/target ratio, and cytokines were monitored at 1-, 3-, 6-, 12-, 18-, 24-, 30-, 36-, 42-, 48-, 60-, and 72-hour time points on a TECAN EVO 100 robotic system. Data are representative of CAR T cells from 1 of 3 individual donors. (B) Luciferase-transduced NALM6 cells (1e6) were injected IV via tail vein into NSG mice on day 0. CAR T cells were injected at day 3, and leukemia growth was evaluated at the indicated time points by bioluminescent imaging. Quantification of bioluminescence at each time point is shown for each individual mouse (bottom graphs, P > .05). Cell surface CD19 CAR expression was evaluated using either a phycoerythrin (PE)-labeled monoclonal anti-FMC63 scFv antibody (Acro) or allophycocyanin (APC)-labeled monoclonal anti-FMC63 scFv antibody. CD22 CAR expression was monitored by staining with a recombinant human siglec-2/CD22 Fc chimera protein (R&D) followed by incubation with a PE- or APC-conjugated goat–anti-human immunoglobulin G (Jackson ImmunoResearch).

Figure 4.

In vitro and in vivo efficacy of CD19-CD22 bicistronic CAR constructs harboring CD28 and 4-1BB costimulatory domains. (A) Schematic representation of the initial bivalent CD19.22.BBζ CAR construct and the newly generated bicistronic constructs, harboring the m971 human anti-CD22 scFv and murine FMC63 anti-CD19 scFv under the control of the EF1α promoter. Constructs differ in the CD28 and 4-1BB costimulatory domains with all hinge-transmembrane domains derived from CD28 in the former and CD8 in the latter. (B) Cytokine production induced by coculture of T cells harboring 1 of the 4 bicistronic CARs or the bivalent CD19.22.BBζ CAR (purple) was evaluated by coculture with CD19⁺CD22⁺, CD19⁺CD22⁻, CD19⁺CD22^High, CD19⁻CD22⁺, and CD19⁻CD22⁻ NALM6 lines. Cocultures were performed at a 1:1 effector/target ratio and cytokines monitored at 1-, 3-, 6-, 12-, 18-, 24-, 30-, 36-, 42-, 48-, 60-, and 72-hour time points on a TECAN EVO 100 robotic system. Results are representative of data obtained in 3 individual T-cell donors. (C) Luciferase-transduced NALM6 cells (1e6) were injected IV into NSG mice on day 0, and the indicated bicistronic CAR T cells were injected at day 3. Leukemia growth was evaluated at the indicated time points by bioluminescent imaging. Quantification of bioluminescence at each time point is shown for each individual mouse (bottom graphs).