Preserved Activity of CD20-Specific Chimeric Antigen Receptor-Expressing T Cells in the Presence of Rituximab

Abstract

CD20 is an attractive immunotherapy target for B-cell non-Hodgkin lymphomas, and adoptive transfer of T cells genetically modified to express a chimeric antigen receptor (CAR) targeting CD20 is a promising strategy. A theoretical limitation is that residual serum rituximab might block CAR binding to CD20 and thereby impede T cell-mediated anti-lymphoma responses. The activity of CD20 CAR-modified T cells in the presence of various concentrations of rituximab was tested in vitro and in vivo CAR-binding sites on CD20(+) tumor cells were blocked by rituximab in a dose-dependent fashion, although at 37°C blockade was incomplete at concentrations up to 200 μg/mL. T cells with CD20 CARs also exhibited modest dose-dependent reductions in cytokine secretion and cytotoxicity, but not proliferation, against lymphoma cell lines. At rituximab concentrations of 100 μg/mL, CAR T cells retained ≥50% of baseline activity against targets with high CD20 expression, but were more strongly inhibited when target cells expressed low CD20. In a murine xenograft model using a rituximab-refractory lymphoma cell line, rituximab did not impair CAR T-cell activity, and tumors were eradicated in >85% of mice. Clinical residual rituximab serum concentrations were measured in 103 lymphoma patients after rituximab therapy, with the median level found to be only 38 μg/mL (interquartile range, 19-72 μg/mL). Thus, despite modest functional impairment in vitro, the in vivo activity of CD20-targeted CAR T cells remains intact at clinically relevant levels of rituximab, making use of these T cells clinically feasible. Cancer Immunol Res; 4(6); 509-19. ©2016 AACR

Figure 1

Rituximab blocks antigen binding of Ab used to derive CAR scFv. Ramos cells (CD20⁺) were incubated with the indicated rituximab concentrations for 30 minutes, followed by incubation with PE-labeled anti-CD20 Ab (clone Leu16) or isotype control at either 4°C or 37°C for 30 minutes. Cells were washed and analyzed by flow cytometry to determine available CD20 binding sites as measured by PE fluorescence intensity. The right panel summarizes the geometric mean fluorescence intensity (MFI) at either 4°C or 37°C as a function of rituximab concentration. The data are representative of 3 independent experiments.

Figure 2

Effect of rituximab on CAR T cell function in vitro. The indicated B-cell NHL cell lines were irradiated and incubated for 30 minutes at room temperature with varying rituximab concentrations (at 2x the concentrations during incubation to yield the indicated final concentrations after addition of T cells). CFSE-stained T cells expressing the Leu16-28-BB-z-tEGFR CD20-specific CAR were added to the target cells at a 1:1 volume and ratio. Proliferation of the T cells was analyzed 4 days later by flow cytometry for CFSE dilution (A). The percent divided CD3⁺ T cells relative to unstimulated T cells are shown on the left axis (filled bars). Histograms of CFSE intensity are shown in Supplemental Figure 2. Cell size of CD3⁺ T cells as determined by geometric mean of forward scatter (subtracting size of cells in media only) is shown on the right axis (open bars). (B) Cytokine secretion of these T cells was measured by Luminex assay using supernatants from 24 hours after restimulation. IL2 concentrations are shown on the left y-axis and IFN-γ and TNF-α on the right y-axis. The data shown are representative of 3 independent experiments.

Figure 3

Effect of rituximab on CAR T cell-mediated cytotoxicity. The indicated ⁵¹Cr-labeled target cells were pre-incubated for 30 minutes with rituximab (at 2x the concentrations during incubation to yield the indicated final concentrations after addition of T cells), and then CD8⁺ T cells expressing the Leu16-28-z CAR were added at the E:T ratios shown in a standard 5-hour ⁵¹chromium-release assay. Mock-transduced T cells, and samples with rituximab and target cells only (“0:1”) were used as negative controls. The average value of duplicate wells is shown, with error bars representing standard deviation. The data are representative of results from 4 independent experiments.

Figure 4

Sensitivity to rituximab blockade is dependent on CD20 antigen density on target cells. K562 cells transduced with CD80 and CD20 (“K80-20”) were cloned by limiting dilution, selected for high, medium, or low levels of CD20 expression (Supplemental Figure 5), and used as target cells in assays for (A) proliferation and cell size (geometric mean forward scatter of gated CD3⁺ cells minus the size of cells in media only) using CFSE-labeled Leu16-28-z CAR-transduced T cells as described in Figure 2; (B) cytokine secretion of the Leu16-28-z CAR-transduced T cells at 24 hours from (A) above, measured by Luminex assay; and (C) cytotoxicity using Leu16-28-z CAR-transduced CD8⁺ T cells by ⁵¹Cr-release assay as described in Figure 3. Data are representative of 3 independent experiments. Absolute values for cytokine secretion are shown in Supplemental Figure 6.

Figure 5

In vivo effect of rituximab on CD20 CAR T cell function. NSG mice were injected i.v. with 5 × 10⁵ rituximab-refractory Raji-ffLuc lymphoma cells, followed by one of the following treatments: no treatment, rituximab only (25 μg or 200) μg i.p 5 days later, 10⁷ 1.5.3-NQ-28-BBz CAR T cells only 6 days after tumor, or rituximab 25 or 200 i.p. at 5 days followed by 10⁷ CAR T cells at 6 days after tumor. Mice were imaged twice weekly for bioluminescence. A) Schema of mouse experiment. B) Average tumor burden per group over time as measured by total body bioluminescence. The geometric mean luminescence values with 95% confidence intervals are shown, and to prevent misleading fluctuations in tumor volume graphs, the last bioluminescence level of each mouse was carried forward after it was killed until no mice in that group remained. Individual bioluminescence traces are shown in Supplemental Figure 8. C) Kaplan-Meier plot showing overall survival of each treatment group. D) Serum rituximab levels on the day of T cell infusion (day 6) and 1 week post T cell infusion (day 13). The red lines denote the median values for each group of mice. E) Serum rituximab concentrations from lymphoma patients who underwent rituximab-containing salvage chemotherapy within the 4 preceding months. The red line indicates the median, and black lines indicate the interquartile range (25-75%).

Figure 6

Effect of ofatumumab on CD20 CAR T cell function in vitro. Irradiated Rec-1 or Raji-ffLuc cells (A and B) or nonirradiated ⁵¹Cr-labeled Rec-1 cells (C) were pre-incubated for 30 minutes with 2x the indicated concentrations of ofatumumab, followed by experiments to determine function of T cells expressing the 1.5.3-NQ-28-BB-z CAR, using the methodologies described in the legend of Figs. 2 and 3. (A) The percent divided CD3⁺ T cells relative to unstimulated T cells are shown on the left axis (filled bars). Cell size of CD3⁺ T cells as determined by geometric mean of forward scatter (subtracting size of cells in media only) is shown on the right axis (open bars). (B) Cytokine secretion of these T cells was measured by Luminex assay using supernatants from 24 hours after restimulation. IL2 concentrations are shown on the left y-axis and IFNγ on the right y-axis. (C) Cytotoxicity of 1.5.3-NQ-28-BB-z CAR T cells was determined using a standard 4-hour ⁵¹Cr-release assay with Rec-1 target cells. The average value of duplicate wells is shown, with error bars representing standard deviation. The data shown each represent a single experiment, with assessment of Rec-1 and Raji targets in independent experiments.