ID #328 Characterizing and optimizing EphA2-targeted CAR T-cells for Group 3 medulloblastoma therapy

Abstract

Chimeric antigen receptor (CAR) T-cell therapy has shown safety and promise in early clinical trials for pediatric brain tumors (PBTs). Medulloblastoma (MB) patients are eligible for CAR T-cell basket trials; however, to-date, there have been no reported CAR T-cell outcomes for MB patients. Further, very few preclinical studies have thoroughly investigated CAR T-cells for MB. In this work, we sought to thoroughly characterize the therapeutic potential and efficacy of EphA2-targeted CAR T-cells for MB. To contextualize our results, we compared EphA2-CAR T-cells with 4H5 scFv and CD28ζ signaling to clinically relevant B7-H3-CAR T-cells with MGA271 scFv and CD28ζ signaling. We show that EphA2 expression is restricted to WNT and Group 3 (G3) MBs whereas B7-H3 is expressed by all MBs. In vitro, EphA2-CAR T-cells have superior cytotoxicity, persistence, and cytokine secretion when cultured against EphA2 and B7-H3 expressing PDX G3MB cell lines (D341, HDMB03, D425). Further, superior in vitro performance of EphA2-CAR T-cells is associated with maintained phenotypic diversity but not reduced exhaustion compared to B7-H3-CAR T-cells. In vivo efficacy of EphA2-CAR T-cells was superior to B7-H3-CAR T-cells in 2/3 cell lines and significantly extended survival of mice bearing D341, HDMB03, D425. Notably, B7-H3-CAR T-cells were superior against D425 which has a high B7-H3 and low EphA2 antigen density. Lastly, we sought to improve EphA2-CAR T-cells by testing 5 different previously published engineering approaches. We show that genetic deletion of DNMT3A or Regnase-1, or expression of a constitutive active IL-18 receptor significantly enhanced in vivo efficacy, while MyD88.CD40 co-stimulation or expression of a constitutively active IL-2 receptor did not. Together, our results support continued clinical development of EphA2-targeted CAR T-cells for G3MB.