The Immunocompromised Mouse

Transplantation of human cancer cell lines or primary human tumors into a host mouse to create a xenograft was one of the first models developed to study PDAC. This approach became possible with the discovery of mice incapable of rejecting the transplanted human cells, first with the T-cell deficient nude athymic mouse, followed by the B- and T-cell deficient severe combined immunodeficient, or SCID, mouse, and the nonobese diabetic, or NOD/SCID mouse. With regards to modeling cancer, the major difference among these mice involves their immune competence and their ability to model the human immune response to cancer. Compared with the SCID mouse, the athymic mouse maintains intact granulocytes, macrophages, B-, NK- and dendritic cell components. Though not often utilized because of expense and technical complexity, NOD/SCID mice can be engrafted with human bone marrow cells to emulate the native immune response.

Cell Line Derived Xenografts

Xenografts can be derived either from established cell lines or directly from human tumors. Cells or tumor tissue can be implanted subcutaneously (heterotopically) or into the mouse pancreas (orthotopically). Each approach has certain advantages and disadvantages. Use of cell lines implanted heterotopically remains a primary tool for pharmaceutical development. This approach is convenient, less expensive, and allows for relatively rapid screening of a large number of candidate molecules to assess efficacy and safety. Unfortunately, this represents the least biologically faithful model. Immortalized cell lines may not emulate the genetic and cellular heterogeneity present in the human disease. Whole exome sequencing studies have shown that there is great genetic diversity, not only among tumors from different individuals, but also within a single tumor and across metastatic sites from the same individual. Although implantation subcutaneously allows for convenient tumor size assessment, growth of the tumor outside of the pancreas and in the absence of tumor stroma does not reflect the human disease; for example, these tumors rarely metastasize. The immunodeficient mouse also is not capable of generating the same immune response seen in human patients. Together, these factors contribute to the outcome that drugs predicted to be effective based on PDAC cell line xenografts often do not translate in the clinical setting.

Patient Derived Xenografts

Patient derived xenografts (PDXs) are generated by transplantation of PDAC tumor tissue directly into mice, either subcutaneously or in the subrenal capsule. PDX tumors may reflect more of the genetic heterogeneity and stromal abundance present in the original tumor. Despite serial passaging in mice, PDX tumors have been shown to maintain some of these features, as well as responses to cytotoxic chemotherapeutic agents. Like standard xenografts, the PDX approach can be used to generate large numbers of mice, serving as a platform upon which therapeutic candidates can be screened. With the advent of gene expression profiling and genomic sequencing, specific gene mutations present in a specific tumor can be studied and response assessed to targeted drugs, administered either as a single agent or in combination. The genetic heterogeneity among PDXs derived from different tumors can be used to model differential responses to chemotherapeutic agents, as seen in patients. For example, a pharmacogenomics model incorporating gene expression profiles and treatment responses was developed with 32 different PDX models of PDAC. These data were used to support a clinical trial further validating the model’s ability to predict treatment response, resistance, and survival in patients with metastatic PDAC.

Many of the disadvantages present in standard xenografts are also problematic in PDXs (eg, an immunodeficient host mouse and subcutaneous tumors implanted distantly from the tissue of origin). Nevertheless, generation and drug screening of xenografts from individual patients has recently been put forward as an approach toward personalized medicine. Overall, this approach is challenging, due to the time and effort required to generate PDXs and to perform drug screening, and the limited availability of effective cytotoxic and targeted agents. The prototypical example of how such an approach could work involves a patient who underwent resection with curative intent for PDAC. The patient’s tumor was used to generate PDXs. Drug screening against these PDXs revealed profound sensitivity to cisplatin and mitomycin C, but relative resistance to gemcitabine. The patient subsequently experienced disease recurrence, which did not respond to gemcitabine. By contrast, when the patient was treated with mitomycin C, a nonstandard approach, the tumor responded dramatically and for a long duration. Ultimately, the patient was found to have a rare germline mutation in the PALB2 gene. Although a terrific example of the potential power of PDX models, it is currently unclear whether such an approach can be applied in patients without uncommon genetic susceptibilities.