Date:February 20, 2018
In our June 2017 blog post, we described advantages and challenges of using syngeneic, GEM, and humanized mouse models for preclinical immuno-oncology (I/O) drug development. In this blog, we expand on this idea and offer thoughts on choosing the most appropriate I/O tumor model for one’s study. While there are benefits and limitations of any model, one can use these considerations, as well as others, as a foundation for preclinical in vivo efficacy study design. Understanding tumor placement, immune composition, response to treatment, and molecular characterization for the model of interest can be invaluable when designing the most appropriate study for your research goals.
The choice of tumor placement has long been important in preclinical development. The classical subcutaneous tumor implant on the flank of the mouse allows for ease of measurement by caliper, is minimally invasive allowing for large studies, and tumor growth kinetics on the flank is generally reproducible. However, subcutaneous tumor models, even those derived from metastatic foci, rarely metastasize1 which create challenges when used for this application. Tumors implanted subcutaneously also lack the organ specific stromal-tumor interactions shown to be important to establishment of the tumor microenvironment. The absence of this interaction can impact both tumor progression and immune response.2
Because of these caveats, orthotopic implantation of both solid and hematological models in preclinical drug development has drawn favor over the past few decades. Some suggest that these orthotopic implants more faithfully represent the clinical situation. For example, several systemic tumor models such as MV-4-11 and C1498 acute myeloid leukemias and MM.1S multiple myeloma, demonstrate seeding in bone marrow which is highly translatable to tumor progression in humans. Additionally, glioma models, such as human U251 and murine GL-261, progress with an invasive phenotype in the mouse brain, very similar to human disease.
As for monitoring tumor progression following an orthotopic implant, the use of bioluminescence imaging (when the luciferase enabled version of the cell line is available) and small animal MRI have a distinct advantage compared to simple survival endpoints determined by clinical observations. The latter manifests clinical signs only in advanced disease, making it very difficult to establish growth kinetics of the tumor under treatment. Such granularity is easily achieved with noninvasive imaging modalities available through MI Bioresearch.
Despite the benefits of orthotopic implantation, a number of challenges are inherent in this strategy. For example, orthotopic models often require surgical placement necessitating technical expertise and constraints in study size. Also, while noninvasive monitoring of tumor progression, such as by luciferase enabled cell lines, is powerful, the generation of these lines may also elicit unintended immune responses that may result in tumor rejection or impact the outcome of immuno-oncology based studies.3 Given the advantages and limitations inherent with subcutaneous and orthotopic tumor placement described here, it is important to rely on the phase of agent development as well as the desired endpoints to dictate the study strategy.
Tumor models vary significantly with regard to immune composition, and knowledge of these relative differences can greatly influence choice of model. These differences are illustrated in Figure 1 for several models offered at MI Bioresearch. Most tumors can be classified as immunologically “hot,” “warm,” or “cold,” most simply defined as the extent to which the immune infiltration of the tumor allows for immune system engagement; “hot” tumors are readily engaging, “cold” tumors are less likely to engage the immune system, and “warm” tumors have elements of each. High and low infiltrates of cytotoxic T cells, respectively, are a major player in this delineation. Likewise, tumors with a high infiltrate of myeloid derived suppressor cells (MDSCs) are typically less responsive to immunomodulatory agents due to enabling an immunosuppressive tumor microenvironment and subsequent immune evasion. Taking this model information, one can hypothesize that a model with a high T cell and low MDSC infiltrate (a “hot” tumor) would be the preferred model to test a checkpoint inhibitor, but this agent is less likely to show activity in a model with low T cell and high MDSC infiltrate (a “cold” tumor). To this end, Figure 2 illustrates the differential response to CTLA-4 checkpoint blockade in the “warm” CT26 model and the “cold” 4T1 model.
Knowing the status of a tumor’s immune cell composition is an important variable for studies designed to target elements of the tumor microenvironment and can allow for powerful targeted combination strategies. For example, the mechanisms of action of radiation and some chemotherapies, as discussed below, can increase activity of immunomodulatory agents in combination that otherwise would be inactive in a “cold” tumor.
Response to Therapy
In the I/O space, the immunogenicity of a tumor can greatly influence response to therapy. As mentioned above, those immunologically “hot” tumor models can respond well to immunomodulatory agents. In contrast, “cold” tumors often do not respond well to immunomodulatory agents for the simple reason that the components necessary for immune system engagement are either not present or are suppressed beyond the ability of the treatment to overcome.
With checkpoint inhibitor therapy being effective only in limited patient populations, these agents are aggressively being employed in combination strategies. Rational study designs using checkpoint inhibitors combined with vaccines, oncolytic viruses, or small molecules are taking center stage in an effort to tip the balance towards stronger patient responses. Preclinically, one of the challenges of combination strategies is identifying a moderate response that allows improvement in combination. This challenge is likely attributed to the plasticity of the immune system in immunocompetent animals. Oftentimes, the variability of response occurs not only study to study, but within a treatment group as well. For example, in a single group with identically staged and treated tumors, one tumor will regress to a durable complete response while the other will evade treatment, growing similarly to control tumors. Such variability in response enables an “all or nothing” paradigm in which the desired outcome of combination therapy is to drive tumors to static or complete responses rather than to slow tumor growth.
While this strategy is appealing, it is more useful with “hot” tumors with T cells ready and waiting for engagement. Targeting “cold” tumors is a more challenging exercise, but is where a strong unmet medical need lies. There is a robust body of literature describing immune system priming with agents that induce immunogenic cell death through enabling an increased pool of antigens for presentation or overall changes to the tumor microenvironment. Chemotherapies such as cyclophosphamide, gemcitabine, 5-flurouracil, as well as radiation, have shown promise in this regard.4,5
Figure 3 illustrates how this approach can be put into practice. 5-flurouracil, anti-PD-1, and radiation each have moderate monotherapy activity in their own right against CT26. In combination, the activities of these agents are enhanced, likely due to alterations of the tumor microenvironment. While additional studies will need to be performed to investigate the mechanism underlying these responses, the power of using model responsiveness to choose combination partners with the best chance of response is clear from these data.
Such conversion from “cold” or “warm” to “hot” tumors as described above has great promise for treatment of cancers such as breast cancer which has shown to be quite refractory to checkpoint inhibitors. As one can see, knowing the relative tumor response and mechanism of a desired combination partner can increase the probability of enhancing immunomodulatory activity of candidate test agents.
One advantage of using syngeneic models in an immunocompetent host is the amassing of over forty years of baseline data. However, molecular characterization in guiding model selection did not become possible until well after these models were developed, and the models enjoyed only limited use over the past few decades as researchers favored access to the human target that xenograft or patient derived models provide. The resurgence of immuno-oncology based research has thrusted syngeneic models back into the limelight, but in the evolution of oncology drug development, these models were woefully lacking in molecular characterization. This deficiency of data is slowly being corrected as groups begin to report genetic analyses across a number of syngeneic models.7,8 However, one should be cognizant that the widespread use of these models across several institutions and over several decades can be accompanied by genetic drift. To this end, rigorous due diligence and literature consensus should be employed to confirm that 1) the cell source being used has the desired mutations in the pathway under investigation, and 2) specific mutations in the model of interest are common across multiple institutions.
In addition to mutational analysis, RNAseq data can also be important to model selection. In situations where a test agent is meant to directly target a tumor, RNAseq data sets can offer insight into whether a target is expressed and to what relative levels. Review of relative RNA expression levels of a target of interest across a panel of models can give greater confidence that the panel being considered is appropriate for testing. This analysis may also assist in correlation of activity to target expression. Taken together, these data will enable one to make informed decisions on the best model or models for use in an in vitro or in vivo study.
Drug development in the I/O space is progressing rapidly, and a wide variety of preclinical models are available to aid in that research. Given the complexities of immunology, one must look beyond cell killing in vitro towards the intricate details of model behavior and immune cell composition. Considerations such as tumor placement, immune composition, response to therapy, and molecular characterization are crucial to the design of a study to address an immunologically based question.
A model, above all else, is a tool. Each of these models has advantages and limitations and should not be used in isolation. Rather, having similar responses of an agent across multiple models with desired characteristics will help increase confidence for success in the clinic. Contact us today to discuss the optimal tumor model for your research needs.
1Hoffman RM (2015) Patient-derived orthotopic xenografts: better mimic of metastasis than subcutaneous xenografts. Nat Rev Cancer, 8:451-452.
2Devaud C, Westwood JA, John LB, Flynn JK, Paquet-Fifield S, Duong CPM, Yong CSM, Pegram HJ, Stacker SA, Achen MG, Stewart TJ, Snyder LA, Teng MWL, Smyth MJ, Darcy PK,4 and Kershaw MH (2014) Tissues in Different Anatomical Sites Can Sculpt and Vary the Tumor Microenvironment to Affect Responses to Therapy. Molecular Therapy 22: 18-27.
3Podetz-Pedersen KM, Vezys V, Somia NV, Russell SJ, and McIvor RS (2014) Cellular Immune Response Against Firefly Luciferase After Sleeping Beauty–Mediated Gene Transfer In Vivo. Hum Gene Ther 25: 995-965.
4Gang C and Emens LA. (2013) Chemoimmunotherapy: Reengineering Tumor Immunity. Cancer Immunol Immunother 62: 203–216.
5Jiang W, Chan CK, Weissman IL, Kim, BYS and Hahn SM (2016) Immune Priming of the Tumor Microenvironment by Radiation. Trends in Cancer 2: 638-645.
6Vincent J, Mignot G, Chalmin F, Ladoire S, Bruchard M, Chevriaux A, Martin F, Apetoh L, Rébé C, and Ghiringhelli F (2010) 5-Fluorouracil Selectively Kills Tumor-associated Myeloid-derived Suppressor Cells Resulting in Enhanced T cell–dependent Antitumor Immunity. Cancer Res 70: 3052-3061.
7Castle JC, Loewer M, Boegel S, de Graaf J, Bender C, Tadmor AD, Boisguerin V, Bukur T, Sorn P, Paret C, Diken M, Kreiter S, Türeci Ö, and Sahin U (2014) Immunomic, Genomic and Transcriptomic Characterization of CT26 Colorectal Carcinoma. BMC Genomics 15: 190.
8Yang Y, Yang HH, Hu Y, Watson PH, Liu H, Geiger TR, Anver MR, Haines DC, Martin P, Green JE, Lee MP, Hunter KW, and Wakefield LM. (2017) Immunocompetent Mouse Allograft Models for Development of Therapies to Target Breast Cancer Metastasis. Oncotarget 8: 30621-30643.
About the Author: Sheri Barnes joined MI Bioresearch in 2017 as Director, Scientific Development. Prior to joining MI Bioresearch, she served as a Study Director at Charles River Laboratories. She has thirteen years of experience in the CRO industry, with the past ten years focused on using in vivo oncology research models in drug development. Sheri holds a Ph.D. in Cell and Developmental Biology from University of North Carolina at Chapel Hill.