AUTHORS
Ryan Urak, PhD, Brenna Gittins, BS2, Huifei Zheng, PhD3, Lior Goldberg, MD, MSc4, Vibhuti Vyas, MS5, Dennis Awuah, PhD6, Ashlie Marie Munoz, PhD7, Stephen J. Forman, MD8, Xiuli Wang, PhD9
AFFILIATIONS
1 City of Hope - Beckman Research Institute, Duarte, California, United States
2 City of Hope - Beckman Research Institute, Duarte, California, United States
3 City of Hope - Beckman Research Institute, Duarte, California, United States
4 City of Hope - Beckman Research Institute, Duarte, California, United States
5 City of Hope - Beckman Research Institute, Duarte, California, United States
6 City of Hope - Beckman Research Institute, Duarte, California, United States
7 City of Hope - Beckman Research Institute, Duarte, California, United States
8 City of Hope - Beckman Research Institute, Duarte, California, United States
9 City of Hope - Beckman Research Institute, Duarte, California, United States
ABSTRACT
Introduction: CD19 CAR T cell therapy has demonstrated remarkable efficacy in leukemia and systemic lymphoma but has shown limited benefit in primary central nervous system lymphoma (PCNSL), with only ~30-40% 1-year overall survival in clinical trials. Limited efficacy may, in part, reflect poor trafficking of intravenously administered CAR T cells across the blood-brain barrier. We previously demonstrated that delivering CAR T cells intraventricularly (ICV) outperformed intravenous delivery. Furthermore, our preclinical studies showed that CD19-CD28 CAR T cells were superior to CD19-41BB CAR T cells following ICV delivery, providing the rationale for an ongoing Phase 1 clinical trial at City of Hope (NCT05625594) using the CD19-CD28 CAR construct.
Methods: To compare, we used our Daudi xenograft intracranial model. In this model, the Daudi cell line (0.2 x 106) was injected into the intracranial space and allowed for 7 days of engraftment. Following engraftment, CD19-CD28 or CD19-41BB CAR T cells (0.5 x 106 for normal dose or 0.3 x106 for suboptimal dose) were injected intraventricularly. Tumor efficacy was analyzed through live mice imaging, and CAR T cell persistence was determined through retro-orbital bleeding. In the rechallenge model, CAR mice were injected subcutaneously with Daudi cells (3 x 106), and mice were imaged to detect relapse. After euthanasia, blood and bone marrow were collected. CAR T cells from blood were isolated and bulk RNA sequenced.
Results: For the first experiment, we treated mice with the optimal dose (0.5 x 106) of CAR T cells. We observed that both CD19-CD28 and CD19-41BB CAR T cells could eliminate the disease. When blood was collected, we observed that CD19-41BB had higher persistence than CD19-CD28. To investigate if these persistent cells had a higher chance of preventing relapses, we repeated the experiment with a second donor and rechallenged the mice at day 35. Interestingly, CD19-CD28 prevented relapse, while CD19-41BB had 4 out of 6 mice relapse. Upon euthanasia, blood and bone marrow were analyzed for CAR persistence, and we still observed CD19-41BB having higher persistence than CD19-CD28. To determine whether CD19-CD28 had higher efficacy, we repeated the experiment but dropped the dose to 0.3 x 106. We observed that while both CAR T cells could treat the disease, CD19-CD28 (8 out of 11 disease-free) outperformed CD19-41BB (4 out of 11 disease-free). To determine if the central nervous microenvironment differentially altered the CAR T cells, we isolated CAR T cells from the peripheral blood of euthanized mice and performed bulk RNA sequencing. We observed CD19-CD28 having higher effector function and metabolism, through expression RPL34, AK3, and MX1 than CD19-41BB CAR T cells, which had increase expression of TIGIT, HMQX1, and DTHD1.
Conclusion: Although CD19-41BB CAR T cells demonstrated greater persistence following intraventricular administration, CD19-CD28 CAR T cells consistently exhibited superior antitumor efficacy, particularly in tumor rechallenge and suboptimal-dose models, indicating that persistence alone does not predict therapeutic outcomes in CNS lymphoma. The enhanced efficacy of CD19-CD28 CAR T cells may reflect differential interactions between CAR signaling domains and the unique CNS lymphoma microenvironment. We hypothesize that the CNS microenvironment limits the effector function of CD19-41BB CAR T cells despite their prolonged persistence. These findings suggest that optimal CAR design may be tissue context-dependent and support the clinical development of intraventricular CD19-CD28 CAR T-cell therapy for CNS lymphoma.
DISCLOSURES
Ryan Urak, PhD: No relevant conflicts of interest to declare.