Abstract Background: Patient-derived Xenograft (PDX) models are being widely used in preclinical studies to identify biomarkers of drug response and to enhance our understanding of cancer biology. Since patients with metastatic cancer have both intra-tumor and inter-site heterogeneity, PDX models generated from different tumor sites may provide a way to study tumor heterogeneity. Characterization of the genomic landscape in these models may also provide better insights into treatment response or resistance. It is rare to have multiple PDX models generated from a single patient over multiple time points during a treatment trajectory. Here, we report the genomic profiles of PDX models generated from 4 distinct tissue specimens over a 7-month period from a patient with metastatic colon adenocarcinoma. The first 2 PDX models were generated from circulating tumor cells (CTCs) and a liver biopsy prior to treatment with a combination pan-AKT + MEK inhibitor regimen. A third PDX model was generated from a liver biopsy while on-treatment and a fourth from an adrenal gland resection at progression. Clinically, all reported metastatic sites, except the adrenal gland, responded to the combination therapy. Results: Genomic characterization of the specimens obtained from these 4 PDX models led to the following observations: 1) PIK3CA E545K and KRAS G12D are present in all the specimens tested for all 4 models and are likely truncal driver mutations; 2) exclusive inter-model SNVs (single nucleotide variants) were identified, and may be model-specific variants representing inter-site heterogeneity in the patient; 3) variants involved in known resistance mechanisms to MEK inhibition were not present in any specimens; 4) overexpression of AKT3 has been reported as a resistance mechanism to a pan-AKT inhibitor and was observed in the adrenal tissue from the patient but not in any other PDX model derived from this patient; 5) intra-model and inter-model heterogeneity in whole genome CNV (copy number variant) profiles was observed between individual PDXs obtained from the pre-treatment CTC-derived model and the on-treatment liver biopsy model. Interestingly, one of the PDXs from the CTC-derived model presented a sub-clonal tumor fraction closely related to the on-treatment liver biopsy model. The multiple inter-model CNV profiles in the liver biopsy derived PDX models represent temporal heterogeneity within a tissue. Conclusions: We observed genomic heterogeneity in PDXs generated from specimens from a patient with metastatic colon adenocarcinoma. Both truncal and sub-clonal variants were identified representing various tumor fractions in these models. This case study illustrates how genomic profiling of multiple tumor sites at different times during course of treatment can provide insight into the complexity of tumor heterogeneity and tumor evolution in patients with metastatic disease. Citation Format: Biswajit Das, Chris Karlovich, Corrine E. Camalier, Rajesh Patidar, Li Chen, Vivekananda Datta, William D. Walsh, Sean P. McDermott, Tomas Vilimas, Palmer Fliss, Justine N. McCutcheon, Amanda Peach, Michelle Ahalt-Gottholm, Carrie Bonomi, Kelly Dougherty, John Carter, Shivaani Kummar, Yvonne A. Evrard, Melinda G. Hollingshead, Paul M. Williams, James H. Doroshow. PDX models generated from a patient with metastatic colon adenocarcinoma display both spatial and temporal tumor heterogeneity [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 1039.