Fratricide-Resistant Anti-CD2 Chimeric Antigen Receptor T-Cells with Endogenous CD2 Knockout Are Highly Effective Against T-Cell Neoplasms

T cell neoplasms, such as peripheral T cell lymphomas (PTCL) and acute T cell lymphoblastic leukemia (T-ALL), have some of the worst prognoses of all hematological malignancies. Poor outcomes are driven by a limited number of effective treatments, particularly in relapsed and refractory settings. While chimeric antigen receptor T cell (CAR-T) immunotherapy has been successful in other hematological diseases, its clinical use against T cell neoplasms is limited due to a lack of unique tumor-specific antigens. Additionally, CAR specificity against pan-T cell antigens causes CAR-T fratricide and insufficient manufacturability for therapeutic dosing. Early studies of CAR-T against CD5- and CD7-expressing T cell neoplasms have shown responses, however, 1) most patients ultimately relapse, and 2) in some patients, antigen negative relapse has occurred. To circumvent these obstacles, we hypothesized that targeting CD2, a pan-T cell antigen, using anti-CD2 CAR-T cells that are engineered to lack CD2, would permit effective CAR-T manufacturing against T cell neoplasms.

In a retrospective analysis of clinical flow cytometry diagnostic samples, we found CD2 expression in 25 of 33 (75.8%) and 37 of 51 (72.5%) pediatric PTCL and T-ALL cases, respectively. Similarly, we found 82 of 89 (92.1%) adult PTCL cases with CD2 immunohistochemical expression, which exceeded CD7 expression (38 of 90, 42.2%) that is currently being evaluated in early phase clinical trials. We optimized early CD2 knockout (KO) during CAR-T cell manufacturing from human donor T cells (KO efficiency: 81.2% ± 2.4, n=9 donors). We next engineered a best-in-class, 4-1BB-costimulated, second-generation CD2KO anti-CD2 CAR-T product (CART2) via lentiviral transduction (PANEL A) using seven distinct single-chain variable fragments (transduction efficiency: 60.2% ± 5.5, n=7 donors) based on cellular expansion and stressed in vivo efficacy against CD2+ T-ALL cells (Jurkat). There were no significant differences in population doublings during CART2 expansion as compared to both CD2 wild-type (CD2 WT) untransduced T cells (Mock UTD) and CD2KO untransduced T cells (CD2KO UTD) (6.20 ± 0.42 vs. 6.12 ± 0.62 vs. 6.67 ± 0.68, respectively, Day 15, n=3 donors), while CD2 WT CART2 cells failed to expand due to fratricide. CART2 cells had no significant differences in T cell memory phenotypes or T cell exhaustion markers, including PD1 and LAG3 expression, at the end of the expansion.

Expanded CART2 cells eradicated primary patient-derived CD2+ Sezary cells and CD2+ T-ALL leukemic blasts as measured by in vitro cytotoxicity assays. Cytotoxicity correlated with CD8+ CART2 cells expressing IL-2 (20.4% vs. 0.2% vs. 0.35%), TNFα (28.5% vs. 0.27% vs. 0.2%), and IFNγ (11.27% vs. 0.18% vs. 0.14%) relative to Mock UTD and CD2KO UTD controls stimulated with CD2+ Jurkat cells (n=2 donors). Importantly, we also found both CD2KO UTD and CART2 cells were capable of TNFα and IFNγ release upon interaction with cytomegalovirus and influenza peptides, suggesting that T cells with CD2 perturbations are still able to mediate effector T cell responses.

We further demonstrated that CART2 cells were highly effective in controlling systemic tumor burden of T-ALL patient-derived xenograft mouse (PDX) models and significantly improved median overall survival as compared to Mock UTD and CD2KO UTD controls (138 days vs. 30 days vs. 30 days, p=0.0007, Mantel-Cox) (PANEL B). Long-term tumor control was proportional to CART2 peripheral blood persistence at Day 28 (CD2KO UTD: 186.6 ± 83.7, n=5 vs. CART2: 936.3 ± 241.7, n=5) and Day 100 (CD2KO UTD: N/A vs. CART2: 514.0 ± 441, n=3). We also found that rescue infusion of CART2 could eliminate relapsed T-ALL in PDX mice previously in a remission after initial treatment with anti-CD5 CAR-T cells, with long-term persistence of CD2-negative T cells thereafter.

In conclusion, our data demonstrate that CART2 are manufacturable and highly efficacious against T-cell neoplasms. To our knowledge, this is the first report of prolonged efficacy using a CD2-targeted cellular monotherapy in human leukemia preclinical models. This product could be delivered alone or in combination with additional CARs for dual targeting, such as anti-CD5 or anti-CD7 CARs, to reduce the potential of antigen-negative escape. Further research is ongoing to study the effect of CD2 deletion in anti-CD19 CAR T cells and the possible implication for clinical translation.