Ex vivo Biomarker Discovery and Validation for Clever-1 Targeting Therapy in Solid Tumors

Abstract

Bexmarilimab is a macrophage-targeting monoclonal antibody that inhibits the scavenger receptor Clever-1 (common lymphatic endothelial and vascular endothelial receptor 1), which is expressed on immunosuppressive macrophages. By blocking Clever-1, bexmarilimab promotes macrophage repolarization toward a pro-inflammatory phenotype. In early-phase clinical trials, bexmarilimab has demonstrated biologically meaningful and clinically measurable anti-tumor activity in a subset of patients with solid and hematological malignancies. However, overall disease control rates have remained modest, highlighting the need for predictive biomarkers and improved patient selection strategies. This thesis builds on a recently developed RT-qPCR based ex vivo stratification method that classifies tumor explants as bexmarilimab-responsive or non-responsive based on treatment-induced gene expression changes. This approach enables predictive biomarker discovery without exposing patients to experimental therapy and allows the use of larger, more representative tumor material than clinical needle biopsies, thereby reducing the impact of intratumoral heterogeneity. Using treatment naïve breast cancer patient-derived explant cultures, we identified lower baseline immune activation across multiple immune-related genes in tumors that exhibited an ex vivo response to bexmarilimab. From an initial panel of 13 immune-related genes, three showed statistically significant differences and were combined into response prediction scores. Clinical validation using pre-treatment needle biopsy samples from a recent clinical trial did not reproduce these baseline expression patterns. This discrepancy is likely influenced by extensive prior therapies in the validation cohort, which are known to substantially alter the tumor immune microenvironment. Together, these findings highlight both the potential and limitations of ex vivo-derived gene signatures and underscore how biological differences between patient populations may hinder clinical translation of predictive biomarkers.