Molecular Profiling and High-Content Drug Screening of Metastatic Colorectal Cancer Organoids Reveal Evolutionary Mechanisms of Pan-KRAS Inhibitor Resistance

Abstract Metastatic colorectal cancer exhibits extensive intertumoral and intratumoral heterogeneity, which underlies variable treatment responses and the inevitable emergence of drug resistance. Preclinical models that accurately reflect this heterogeneity and enable the investigation of resistance mechanisms at the clonal level, while also providing sensitive and quantitative measurements of treatment response, remain limited. Here, we describe a living biobank of patient-derived organoids (PDOs), generated from metastatic colorectal cancer samples encompassing various genetic backgrounds, including different KRAS genotypes. These PDOs retain key phenotypic, genomic, and transcriptional features of their matched tumors, and, when coupled with a multiparametric high-content 3D drug-screening platform integrating volumetric growth, nuclear content, and proliferative readouts, enable sensitive detection of subtle responses that are not captured by conventional viability-based assays. Systematic phenotypic profiling identified KRAS mutational status as a major determinant of drug response, with KRAS wild-type organoids exhibiting predominant cytotoxic collapse, whereas KRAS-mutant PDOs displayed cytostatic growth arrest. Building on this phenotypic stratification, we investigated resistance to the pan-KRAS inhibitor BI-2865 by integrating longitudinal drug exposure with expressible DNA barcoding and single-cell RNA sequencing to resolve clonal and transcriptional dynamics. Resistance to BI-2865 arose primarily through transcriptional plasticity rather than widespread genomic alterations. Lineage tracing revealed genotype-dependent trajectories, with de novo emergent clones driving resistance in KRAS WT PDO, whereas KRAS G13D models were dominated by expansion of pre-existing clones. Single-cell analysis on KRAS G13D demonstrated a shared transcriptional landscape comprising multiple resistant states rather than a single convergent identity. Together, our findings define resistance to pan-KRAS inhibition as a modular, state-resolved process driven by transcriptional plasticity and genotype-dependent clonal dynamics and establish PDO-based multiparametric phenotyping combined with lineage-resolved single-cell profiling as a robust framework for dissecting the evolutionary framework of therapeutic resistance in metastatic colorectal cancer.