Abstract Activating mutations of the KRAS gene are found in ~90% of pancreatic (PDAC) and ~50% of colorectal (CRC) cancers. However, therapeutic strategies designed to counter or extinguish KRAS signaling remain unsuccessful. Redundant signaling pathways and a complex solid tumor microenvironment are key barriers to current KRAS targeting efforts. Most conventional drug testing targeting KRAS has commonly used conventional cell line systems that lack clinical relevance and biological heterogeneity and thereby failed in patients. More recently, patient-derived models, including primary cell cultures, organoids or spheroids have been deployed in large screens designed to identify effective combination regimens. Since patient-derived models more clinically representative, lead single agents or combinations are exhibiting a higher likelihood of translational success. Our approach long-term goal is to establish and test agents using patient-derived xenografts (PDX) coupled with an ex-vivo tumor tissue culture assay that more effectively tests drug sensitivity. Using this ex-vivo method we tested 36 drug combinations against a panel of 10 PDAC PDX tumors to identify drug combinations that demonstrate synergistic effects. These combinations include, MEK inhibitors along with multi-kinase inhibitor, suntinib, or CDK4/6 inhibitors. Synergistic anticancer activity using MEK-CDK4/6 combinations also occurs in our CRC PDX models that have translated to an ongoing clinical trial. Our preliminary studies and ongoing clinical trial provide a strong rationale to further test these combinations in large cohorts of PDXs. In the proposed study, we aim to: 1) determine the effectiveness of the MEK inhibitor-based combinations through in vivo PDX trials; 2) establish and characterize PDXs from patients enrolled in a clinical trial that combines MEK and CDK4/6 inhibitors; and 3) identify biomarkers of responding tumors and explore the mechanisms of resistance. The identification of lead combinations in studies using clinically relevant patient-derived model systems predicts a higher chance of success in future clinical studies. In addition, all proposed test agents are approved for use against cancer or are being evaluated in the clinical trials. Thus, there is a high probability that these efforts will be quickly translated to the clinic and rapidly benefit PDAC and CRC patients.