A good model system that closely mimics the in vivo tumor environment is required to guide clinical decision-making and evaluate novel therapies for pancreatic cancer.
As no one model succeeds in mimicking the human pancreatic tumor environment, the Hale Family Research Center employs multiple model systems to capture all facets of the tumor environment of a patient.
Cell line. The ideal cell model system should represent a broad variety of cancer subtypes; have an intact tumor microenvironment, including stroma, extracellular matrix, and immune cells; and be readily able to propagate at reasonable cost, while maintaining the heterogeneity and complexity of the original tumor. Although several models have been used and some continue to be developed, they are all characterized by inherent pros and cons.
Cell lines, grown on plastic dishes, have been the traditional approach. Cell lines are homogeneous and can be grown indefinitely in culture but suffer from limited availability for any given cancer type (therefore do not represent the full spectrum of genetic alterations), lacking in structural organization, heterogeneity, and an intact tumor microenvironment.
Organoid. Tumor organoids, cultured three-dimensionally (3D) in the presence of matrigel (a gelatinous protein mixture that mimics the extracellular matrix found in many tissues), are typically derived from fine-needle aspiration or surgical resection.
Advantages of 3D growth include that organoids are established at higher success rates from patient material than cell lines, preserve heterogeneity of the original tumor, and maintain cell-cell and cell-matrix interactions more closely than traditional 2D growth on plastic.
Patient-derived xenografts (PDX). While organoids offer some advantages in representing the native microenvironment, they lack stroma and an intact immune environment. Patient-derived xenograft models (PDXs) are derived by direct implantation, either subcutaneously or orthotopically, of patient material into immunodeficient mice, which leads to tumor formation and subsequent passage of the tumor in mice.
PDX models are believed to most closely mirror the drug responses that are observed in patients. Their strengths include an intact tumor microenvironment and at least certain elements of the immune system that are preserved. However, PDXs are the costliest models, require lengthy passaging and a lot of tissue for establishment, and they cannot be used for large-scale screens, limiting their utility to confirmation of results obtained in other systems.
Genetically engineered mouse models (GEMMs). Genetically engineered mouse models (GEMMs) of pancreatic cancer closely mimic the human disease. The mice that we use, referred to as the Lox-Stop-Lox (LSL)-KrasG12D; LSL-Trp53-/-; Pdx1-cre (KPC) model, involve combined loss of the p53 tumor suppressor and conversion of KRAS into its oncogenic form only in pancreatic tissue, since Cre expression is driven from a pancreas-specific promoter.
Pancreatic tumors are generated with high penetrance in a microenvironment similar to human tumors, and since immunocompetent mice are used, immune-modulatory agents (for example, immune checkpoint inhibitors) can be tested. Despite these advantages, there are drawbacks, since GEMMs are relatively expensive and time consuming to establish and maintain.
Models for direct patient impact
The goal is in the near future to have established a living biobank consisting of tumor organoids, which can retrospectively be used to identify therapeutic strategies tailored to specific genetic alterations.
We hope to establish organoids from patients and screen these in “real time” to identify specific therapeutics, or combination of therapeutics, that will be relevant for treatment of that patient, thus allowing us to have a direct and immediate impact.
Tumor organoids represent an intermediate model between cancer cell lines and PDXs. While this new technology holds much promise, it remains to be seen if the therapeutic predictive utility of organoid models is superior to other models and whether they can be used to identify novel diagnostic strategies for patients.