Optimizing Therapeutic Outcome of CD19(T2)28z1XX Chimeric Antigen Receptor (CAR) T in Adult Patients with Large B-Cell Lymphoma through a Quantitative Medicine Approach

• Published in: Blood 2024-11, Vol.144 (Supplement 1), p.7138-7138
• Main Authors: Dey, Agnish, Li, Jia, Park, Jae H., Sellner, Leopold, Sundaresan, Varsha, Palomba, Maria Lia, Mansilla-Soto, Jorge, Senechal, Brigitte, Wang, Xiuyan, Sikder, Devanjan S, Dash, Ajeeta, Riviere, Isabelle, Sadelain, Michel, Mugundu, Ganesh
• Format: Article
• Language: English
• ISSN: 0006-4971; 1528-0020
• DOI: 10.1182/blood-2024-203375

Background: CD19(T2)28z1xx is an autologous CD19-targeting CAR-T cell therapy consisting of CD28 and optimized 1XX CD3ζ signaling domains with the hypothesis being 1XX CAR-T cells are more potent and will display promising efficacy profile with a lower infusion dose. Non-linear mixed effects modeling was used to characterize the cellular kinetics (CK). The developed CK model was further leveraged to assess the impact of various patient characteristics on CD19(T2)28z1xx kinetics. Exposure-response (E-R) relationships were evaluated for safety (probability of high-grade (>1) cytokine release syndrome (CRS)) and efficacy (rate of overall response (ORR) and complete response (CR)) to support dose recommendation. Methods: Analysis presented here was performed using data from the clinical study for four dose levels (25 x 106, 50 x 106, 100 x 106, 200 x 106 CAR-T cells). For E-R efficacy analyses, ORR and rate of CR were evaluated across all dose levels against exposure metrics (Cmax and AUC) using logistic regression. Similarly, for E-R safety analysis, rate of high-grade CRS was evaluated against the exposure metrics. Population CK model was developed using piecewise functions embodying an initial cellular expansion followed by bi-phasic contraction. A forward selection and backward elimination stepwise covariate modelling approach was used to add covariates to the model. The focus of covariate modelling was to identify parameter-covariate relationships that can potentially explain the inter-individual variability observed with CK data. Results: The exposure-response analyses for ORR indicated that the probability of ORR was around 85% and probability of CR was around 74% across all dose levels and observed range of exposure. The relationship between exposure metrics and efficacy was flat. While the relationship between the incidence of high-grade CRS and exposure was not deemed statistically significant, a higher likelihood of experiencing high-grade CRS was noted with an increase in exposure. Therefore, the lowest dose level of 25 x 106 CAR-T cells was selected as the recommended phase 2 dose (RP2D) for CD19(T2)28z1xx. Developed population CK model was able to effectively characterize multiphasic kinetics estimating parameters such as doubling time, initial decline half-life, and terminal half-life (representing long-term persistence). A covariate search of forward addition and backward elimination on the population CK model indicated that ECOG (Eastern Coope