Abstract 855: Model-informed drug design (MIDD) exploration of a theoretical CD47xTAA bispecific antibody using available data from magrolimab, an anti-CD47 monoclonal antibody

Abstract

Abstract Purpose: The goal of this study is to outline the process and utility of using fit-for-purpose, semi-mechanistic PK/PD modeling to inform the design of a theoretical CD47 bispecific antibody targeted to a generic tumor-associated antigen (CD47xTAA bsAb) with improved efficacy, tolerability, and PK compared to single target anti-CD47 therapies. Approach: To help guide the CD47xTAA bsAb design, a 3-compartment human model was built to include a target-mediated drug disposition (TMDD) component from CD47 expressed on red blood cells (RBC), white blood cells (WBC), and cells in the peripheral compartment. The model was parameterized using available literature data and calibrated against magrolimab human PK data that exhibited potential TMDD. The magrolimab model was used to identify key CD47 receptor occupancy (RO) thresholds associated with tolerability (the priming dose threshold) and efficacy (the recommended Phase 2 dose/RP2D threshold). Next, assumed drug properties and TAA expression levels for a CD47xTAA bsAb were implemented in a similar human model and drug PK and doses required to match key CD47 RO thresholds were explored across a range of single-arm binding affinities. Results: The calibrated magrolimab model projected target engagement levels on RBCs and WBCs and showed rapid antibody-induced CD47 pruning on RBCs that is consistent with clinical PD measurements. At TAA binding affinities that produce sufficient binding to tumor antigen, the model predicts that lowering the CD47 binding affinity allows for a higher starting dose that matches the priming threshold, while lowering the dose required to match the RP2D threshold. The model also shows that lowering the binding affinity to CD47 may improve PK exposure in the tumor compartment by reducing systemic CD47 TMDD. Conclusions: This modeling exercise demonstrates how semi-mechanistic PK/PD models can be used in model-informed drug design (MIDD) by leveraging the targeting advantage of avid bispecific antibody therapies and by using publicly available preclinical and clinical data. Furthermore, these model projections can help inform future studies with bispecific CD47 drug candidates. Citation Format: Emily A. Pace, Bjorn L. Millard. Model-informed drug design (MIDD) exploration of a theoretical CD47xTAA bispecific antibody using available data from magrolimab, an anti-CD47 monoclonal antibody [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 855.