The Effects of Co-Stimulatory Endodomains on the Fate of T Cells Expressing a Tumor Directed Chimeric Antigen Receptor (CAR) In Human Subjects with B Cell Malignancies

Abstract 3949

B cell lymphomas have been effectively treated by immunotherapy, including monoclonal antibodies (MAbs) and adoptive T cell transfer. To extend this success, investigators have genetically modified T cells to express a B cell specific antibody incorporated in an artificial chimeric antigen receptor (CAR), essentially combining antibody and cell-based approaches. Early B cell directed CARs combined the antigen binding domains of the variable regions of a CD19 or CD20 MAb (scFv) with the CD3ζ endodomain of the TCR/CD3 complex. Although such CARs confer potent cytotoxic function to T cells, initial clinical trials showed that T cells modified to express CARs engineered in this way had limited in vivo persistence, apparently receiving insufficient costimulation following CAR engagement.

To overcome the above limitations, a multiplicity of costimulatory endodomains, including CD28, 41BB or OX40, have been incorporated into the CAR molecule. Because of patient heterogeneity, it has proved difficult to draw definitive conclusions about the relative expansion, persistence and effectiveness of cells with each modification, so that their comparative value in human subjects remains speculative. We therefore designed a phase I, dose escalation clinical trial in which patients with refractory/relapsed B cell malignancies were simultaneously infused with two autologous T cell products. Both express a CAR with an identical CD19-specific exodomain, but one CAR also has a CD28-ζ endodomain while the other expresses only a ζ endodomain. With this study design, each patient acts as a “self-control”, allowing us to directly discover the consequences of CD28 costimulation for the fate of the T cells in vivo.

We enrolled two patients at each of the three cell dose levels (1^st level = 2×10⁷/m² of each product; 2^nd level = 1×10⁸/m²; 3^rd level = 2×10⁸/m²). End points of the study were safety, persistence of each of the two generations of CAR-modified T cells, and assessment of antitumor activity. T cell products were generated by activation of autologous PBMC with immobilized OKT3 and gene modified with retroviral vectors encoding either CAR.19ζ or CAR.19-28ζ. After transduction, T cells were expanded ex vivo for a median of 14 days (range 6–18) in the presence of IL-2. CAR expression was 42%±18% and 49%±16% for CAR.19ζ and CAR.19-28ζ, respectively. This corresponded to 51,246±16,795 and 18,283±9,484 transgene copy numbers/μg DNA, respectively, as measured by Q-PCR. Products contained both CD8⁺ cells (CAR.19ζ = 49%±22%, CAR.19-28ζ = 48%±22%). Few naïve T cells were present in either transduced population (CD45RA⁺ = 6%±5% and 6%±6%, respectively), and memory T cells predominated in both (CD45R0⁺CD62L⁺ = 50%±24% and 47%±66%, respectively). Both T cell components equally and specifically targeted CD19⁺ tumors in vitro as assessed by ⁵¹Cr release assays (specific lysis was 53%±10% for CAR.19ζ and 65%±19% for CAR.19-28ζ at a 20:1 E:T ratio). All infusions were well tolerated in all patients. Persistence of CAR⁺ T cell was assessed in peripheral blood by Q-PCR assays specific for CAR.19ζ and CAR.19-28ζ.

Molecular signals for CAR.19-28ζ began at a low level after infusion, but progressively increased (7 to 63 fold) to peak at 1–2 weeks post infusion, before declining to background levels over the ensuing 8 to 13 weeks. By contrast, molecular signals corresponding to CAR.19ζ⁺ cells were barely detectable after infusion, showed no expansion, and rapidly disappeared. Currently 4 patients are evaluable for disease response; 2 had stable disease for up to 6 months and 2 had progressive disease.

Hence, infusion of both CAR.19ζ and CD19-28ζ T cells has been safe at current doses. Direct comparison of each cell product in individual patients indicates that inclusion of the CD28 costimulatory endodomain (2^nd generation CAR) enhances expansion and persistence. Nonetheless, both the limited expansion and persistence and the modest clinical effects suggest that additional modifications will need to be made to CAR endodomains to optimize the benefits of this therapy. We suggest our approach will allow these modifications to be evaluated systematically and directly even in small-scale clinical studies.