CAR-T with ICOS costimulation lacks therapeutic efficacy in patient due to strong trogocytosis and failure of expansion

Abstract Background CAR-T cell therapy targeting B-cell antigens has become an established strategy for relapsed hematological malignancies. However, development of even more effective CAR-T therapy to overcome antigen-escape relapse is warranted. We previously demonstrated that dual targeted CD19/CD20-4-1BB CAR-T therapy was highly effective (Nat Med, 2020). To further improve outcome, we generated trispecific duo CAR-T cells (CD19/CD20/CD22) with tandem anti-CD20/CD19 CD3z/ICOS intracellular signaling domains (ICD) and a separate anti-CD22 scFv with CD3z alone. This construct exhibited efficient T-cell signal transduction and tumor control in vitro and in NSG mice (Sci Trans Med, 2021). Surprisingly, trispecific-ICOS CAR-T cells failed to expand in patients and thus lacked therapeutic efficacy in a Phase I (NCT05094206) clinical trial (ASH 2024). In this study, we examined the potential mechanisms underlying the failed response of trispecific-ICOS CAR-T cells. Methods Based on our IRB approved protocol and using patient-derived PBMCs and leftover samples from treated patients, we generated tested bispecific-4-1BB, bispecific-ICOS and trispecific-ICOS CAR-T cells, and tested their activation and function. In vivo xeno-graft mouse models were used to better understand CAR kinetics, response, and outcomes. Results We first examined the ability of CAR-T cells to control Raji cell growth in an in vivo mouse model using a lower, more clinically relevant, dose than what had been used in prior studies. Recipients of bispecific- or trispecific-ICOS CAR-T cells showed worse survival compared to those of bispecific 4-1BB CAR-T cells. Similarly, ICOS CAR-T cells failed to expand and exhibited an exhaustion phenotype in Raji-bearing mice. Since ICOS and 4-1BB CAR-T cells expanded similarly during manufacture with IL-7/IL-15 cytokines, we hypothesized that ICOS CAR-T cells may be defective in their ability to response to antigen. To test this, we co-cultured CAR-T cells with Raji cells for 1, 5, 15 and 40 hours, and found that these CAR-T cells quickly down-regulated CAR expression as reflected by loss of surface FMC63 staining. Given that total FMC63 levels were not reduced in any type of CAR-T cells through surface and intracellular staining, these results indicate that CAR was internalized. Strikingly, ICOS CAR-T cells significantly enhanced surface-FMC63 loss compared to 4-1BB counterparts after culture for 1, 5 and 15 hours. Surface-FMC63 expression was mostly recovered at 40 hours on 4-1BB CAR-T cells, but not on ICOS CAR-T cells. Trogocytosis, a process whereby some proteins from one cell are transferred to another cell, has been reported to occur between CAR-T cells and their targeted tumor cells contributing to antigen loss on tumor cells and fratricide CAR T-cell killing. We thus examined CD19 expression on CAR-T cells after Raji stimulation and observed transient expression of CD19 on 4-1BB CAR-T cells, suggesting that low level trogocytosis was occurring. Surprisingly, ICOS-CAR T cells demonstrated higher degree and more sustained surface CD19 expression as well as increased apoptosis, suggesting augmented fratricidal T-cell killing. To exclude the impact of allogeneic responses to Raji cells, we purified B cells from PBMCs derived from the same patient as the CAR-T cells. Consistent with the Raji cells, ICOS-CAR-T cells showed significantly decreased surface-FMC63 and increased apoptosis compared to 4-1BB CAR-T cells after co-culture with autologous B cells. Taken together, these data indicate that ICOS ICD enhanced trogocytosis and fratricide of CAR-T cells. We further co-cultured CAR-T and Raji cells for 2 weeks, to mimic the in vivo condition where CAR-T cells are continuously stimulated by lymphoma cells in patients. While 4-1BB CAR-T cells expanded during the culture period, ICOS CAR-T cells declined in numbers. ICOS CAR-T cells exhibited decreased proliferation (Ki67) and increased markers of exhaustion (Tim3, PD-1 and Lag3). These results indicate trogocytosis as the key mechanism underlying failed expansion and lack of response of ICOS CAR-T cells in our clinical trial. Conclusions In conclusion, we found that ICOS ICD signal drives strong trogocytosis and promotes surface expression of tumor antigen that prevents CAR-T cells from expanding properly. Thus, our findings raise significant caution for using CAR constructs that rely on ICOS for CAR activation.