Translational modeling for patients with RAS mutant tumors: Profiling the dual-MEK inhibitor IMM-1-104 in a humanized 3D assay.

e15084 Background: Elevated RAS-RAF-MEK-ERK (MAPK pathway) signaling is observed in over half of all solid human tumors, and mutations in RAS or RAF account for a large fraction. Given MEK’s unique position in the MAPK cascade, it remains an attractive target in cancer. However, FDA-registered MEK inhibitors are susceptible to pathway reactivation events that limit their use to RAF mutant disease and cause on-target toxicities stemming from chronic target engagement. IMM-1-104 is a novel, allosteric dual-MEK inhibitor designed for better applicability to RAS mutant tumors by preventing MEK reactivation. Endowed with a short plasma half-life, IMM-1-104 promotes deep cyclic inhibition with a near-zero drug trough, affording normal cells a chance to recover between doses. Methods: We characterized IMM-1-104’s pharmacologic activity across 52 tumor cell lines that spanned 11 distinct tumor types in a humanized, ECM-based 3D tumor growth assay (3D-TGA). The 3D-TGA has better predicted in vivo tumor responses versus 2D culture and more accurately reflects human tumor biology. Tumor models were categorized based on in vivo drug PK limits as sensitive to IMM-1-104 (EC50 < 1uM), intermediate (1uM≤EC50≤10uM and ≥25% inhibition at 10uM) or resistant otherwise. Models were evaluated by whole exome sequencing, along with RNA sequencing in the 3D context, to profile determinants of sensitivity and resistance and to prioritize patient populations most likely to respond to IMM-1-104. Results: Models sensitive to IMM-1-104 were enriched for MAPK driver mutations, consistent with pathway addiction. We reasoned that activation of parallel compensatory pathways that can reduce reliance on MAPK signaling may increase the likelihood of resistance to IMM-1-104. Pathways and genes suspected of contributing to resistance helped refine signatures based on 3D-TGA outcome data. Models with a MAPK driver mutation and compensatory mutations such as PIK3CA or PTEN deletion were more likely to show intermediate response than those with a greater addiction to MAPK drivers. Models lacking a clear MAPK driver mutation but harboring other putative resistance alterations were more likely to be resistant in the 3D-TGA. Conclusions: To better understand the relevance of tumor model responses in the 3D-TGA relative to RAS mutant patient populations, we computationally compared tumor model data to patient somatic alterations, identified in the public resource GENIE, which has cataloged the molecular profiles of over 100,000 cancer patients. Based on model-to-patient molecular mapping, we identified biomarker-defined subsets of sensitive KRAS mutant lung and colorectal models. The most broadly sensitive patient-aligned models in the 3D assay were KRAS mutant pancreatic cancer and NRAS mutant melanoma patients, supporting the inclusion of such patients in planned clinical studies of IMM-1-104.