Abstract 4705: Late ALL relapse following CD19 CAR immune-pressure demonstrates reversible pan-antigen loss

• Published in: Cancer research (Chicago, Ill.) Ill.), 2015-08, Vol.75 (15_Supplement), p.4705-4705
• Main Authors: Jacoby, Elad, Yang, Yinmeng, Chien, Chris D., Haso, Waleed, Qin, Haiying, Fry, Terry J.
• Format: Article
• Language: English
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.AM2015-4705

Despite remarkable remission rates in relapsed and refractory acute lymphoblastic leukemia (ALL) following CD19 chimeric antigen receptor (CAR)-T cells or CD3-CD19 bispecific constructs, relapse of CD19 negative leukemic blasts occurs in approximately 10-20% of cases. The mechanism of antigen loss under CD19-targeted immune pressure on pre-B ALL has yet to be elucidated. Although xenograft models have been excellent tools for pre-clinical studies, the development of xenogeneic GVHD has precluded the ability to study late leukemic relapse under CD19-targeted immunotherapeutic pressure. Thus, we performed such studies in a murine transplantable pre-B ALL (CD19+ B220+ CD22+ sIgM- CD127+ CD43+) in immunocompetent mice that induces lethality of recipients within 15-25 days without therapy. Treating ALL bearing mice with a murine CD19 CAR T cells (CART), resulted in prolonged remissions followed by CART dose-dependent relapses varying between 50 days to >200 days post treatment. All relapses were negative for CD19 expression with earlier relapses maintaining similar pre-B cell surface marker phenotype of the parent ALL. However, relapses longer than 100 days consistently showed a dramatic alteration in phenotype, with loss of CD22, B220 and CD127, stability of CD45 and CD43, and partial gain of cKIT, resembling a more immature B cell or common lymphoid progenitor. While early relapses had normal mRNA levels of CD19 despite loss of surface antigen, late antigen loss correlated with low CD19 mRNA, along with other alterations in pre-B cell signature on microarray. This late antigen negative leukemia took longer time to engraft in secondary recipients, compared to the parent cell line. However, serial transplantation without immune-pressure resulted in regain of CD19 and CD22 antigens with tertiary recipients demonstrating leukemia phenotype and time to engraftment similar to the parent cell line. Interestingly, in-vivo reacquisition of CD22 always preceded regain of CD19. To further study loss of B cell antigen, we co-cultured CAR T cells with both human and murine leukemia. Using CD19 CART, we observed rapid CART dose-dependent loss of surface CD19 followed by partial loss of surface CD22, but not other leukemia markers, in both systems. Interestingly, antigen loss was not seen when co-culturing leukemia with anti-CD22 CART. Suboptimal dose of CD22 CART in an NSG mouse model of human leukemia did generate a CD22-CD19- late relapse. We conclude that loss of CD19 m