05/23/2024

In silico tumor immune microenvironment (TiME) analysis of non-small cell lung cancer (NSCLC) to inform clinical development of CDR404: A first-of-its-kind MAGE-A4 targeted T-cell engager

ASCO 2024 Abstract
Authors Daniel Lenherr-Frey, Melissa Vrohlings, Maria Bouvy-Liivrand, Reija Hieta, Heather May Shaw, Gilberto Lopes, Leonardo Borras, Swethajit Biswas

Background: CDR404 is an antibody-based bivalent MAGE-A4 targeted T-cell engager (TCE). One key mechanism-of-action of TCEs is CD8 T-cell redirection which involves T-cell intravasation into tumors [Damato et al, 2019]. CDR404 mediated TiME remodeling will likely be dependent on the inflammatory (INFLAM) and vascular (VASC) phenotype especially since tumor vasculature constitutes functional and physical barriers to T-cell infiltration [Sahu et al, 2022] [Desbois et al 2020] [Duru et al, 2020]. To identify biomarkers for CDR404 anti-tumor responses in NSCLC, we evaluated the associations between MAGE-A4 mRNA expression, immune cell populations and frequency of 9p21 deletions mediating T-cell infiltration [Han et al, 2021].

Methods: Expression of MAGE-A4 mRNA was evaluated by the Tempus xR RNA-Seq assay (Tempus AI, Inc. Chicago IL). ConsensusTME bulk RNA-Sequencing deconvolution [Jiménez-Sanchez et al, 2019] was used to analyze 16 TiME cell lineages in two NSCLC TCGA datasets – primary LUSC & LUAD [Giacomazzi et al, 2023]. Immune cell data was stratified by MAGE-A4 expression quartiles (Q): Null = not detected; MAGE-A4LOW = Q1-Q3; MAGE-A4HIGH = Q4. GISTIC2.0 was used to identify 9p21 gene deletions [Mermel et al, 2011].

Results: MAGE-A4 is enriched in LUSC. MAGE-A4 levels were similar across metastatic organ sites (e.g., liver vs. lymph nodes) in LUSC indicating that tumor location is not a confounding factor for TiME analysis. In LUSC, MAGE-A4HIGH vs. MAGE-A4NULL tumors had lower levels of 15/16 immune cell populations. Largest reductions were in endothelial cells (p=7.71-e05) and CD8 T-cells (p=0.00096). Deletions in 9p21 genes were more frequent in MAGE-A4HIGH vs. MAGE-A4NULL tumors, e.g., CDK2NA: 91% vs. 68% (p=0.0028), consistent with reduced CD8 T-cells in MAGE-A4HIGH. In MAGE-A4HIGH vs. MAGE-A4LOW reductions in 4/16 immune cell populations were seen. In LUAD, immune cell levels were similar across MAGE-A4 subgroups except lower endothelial cells in MAGE-A4HIGH vs. MAGE-A4NULL (p=0.00055).

Conclusions: LUSC MAGE-A4HIGH tumors had a differentiated TiME profile. Our findings are consistent with an INFLAMLOWVASCLOW phenotype possibly indicative of an “immune desert” [Desbois et al 2020]. In contrast, LUAD MAGE-A4HIGH tumors had an INFLAMHIGHVASCLOW phenotype indicating that MAGE-A4 associations with TiME may be histology dependent in NSCLC. Overall, in MAGE-A4HIGH LUSC & LUAD, susceptibility to CDR404 mediated T-cell tumor intravasation may be better because of a lower angiogenic barrier. Translational baseline tumor biopsy sub-studies from the CDR404 Phase 1 trial are awaited to confirm if INFLAM/VASC phenotype is predictive of response in relapsed locally advanced/metastatic NSCLC patients. CDR-Life acknowledges Tempus AI, Inc. for the expression analysis using the Tempus xR RNA-Seq assay.

VIEW THE PUBLICATION