Introduction

Mithun Khattar1 and Karuppiah Kannan2

1Obsidian Therapeutics, Inc., 1030 Massachusetts Avenue, Cambridge, MA 02138, USA

2Takeda Development Center Americas, Inc., Cambridge, MA 02139, USA

The ability of our immune system, to specifically recognize and kill cancer through Immune-surveillance, is a phenomenon known to mankind for over a century. It was in 1890s when Dr. William Coley, now known as father of cancer immunotherapy, could achieve long-term remissions from disease in several sarcoma patients treated using live and killed strains of bacteria (Coley 1891, 1893). At that time, there was only a basic understanding of the immune system and none of the immune cells had been discovered. Based on Dr. Coley's findings, Dr. Paul Ehrlich postulated that "Cancer would be quite common in long-lived organisms if not for the protective effects of immunity" (Ehrlich 1909). However, it was only recently that the concept of Immune surveillance of tumors was established in preclinical models, with evidence that mice lacking the adaptive immune system are more susceptible to developing tumors. Tumors in these mice grow more rapidly compared to wild-type mice, and do not respond to therapies that reinvigorate the host immune responses (Shankaran et al. 2001; Matsushita et al. 2012). Similar findings have been reported in humans, suggesting increased susceptibility to cancer in recipients of organ transplants who undergo systemic immune-suppression or otherwise immune-suppressed individuals (Engels et al. 2011). A striking finding around the same time that solidified the immune-surveillance concept was that cancers evade immune recognition by harnessing the immune-checkpoint (e.g. PD1/PDL1, CTLA4, TIM-3, LAG-3, etc.) pathways, which act like a brake on the endogenous antitumor immune responses (Korman et al. 2006). Indeed, therapies using interleukins, interferons as well as stem cell transplants have been practiced in the clinic since 1970s to treat various types of malignancies, basically engaging or assisting the immune system to mount an endogenous antitumor immune response.

Subsequently, within the last decade, the curative potential of cancer immune therapies including check-point blockade and adoptive T-cell therapy was demonstrated in a subset of patients and indications. Inevitably, these breakthrough clinical successes have led to exponentially growing campaigns on the search for next-generation drugs aimed at educating and activating the immune system to recognize and kill cancer. This quest is expected to result in "cures," a term cancer researchers and clinicians for the first time in history are comfortable acknowledging in a wide variety of oncology indications. With this enthusiasm, there are currently more than 1000 cancer immunotherapy agents being tested in clinical trials and at least 3000 in preclinical development (an under-estimate due to undisclosed agents by several companies) (Tang et al. 2018). With the likelihood of approval (LOA) of drugs entering clinical trials being lowest in oncology (<5%) among all disease indications, accurate and relevant preclinical modeling of these cancer immunotherapy agents is a highly sought-after holy grail. Murine models are widely used for preclinical evaluation of all drugs owing to their ability to grow tumors rather uniformly and reproducibly in a short duration, allowing performance of large cohorts of studies under controlled conditions, which is beneficial for the time-sensitive and competitive nature of preclinical development. While these attributes are desirable for logistics and advancement of preclinical development, the same along with other factors can be the potential reasons for translational failures, i.e. lack of relevance in patients. The overarching goal of this book is to not only provide the readers with a baseline understanding of different preclinical model systems available for development of novel cancer immunotherapies, but also highlight aspects of these individual models that do and do not translate into clinical findings. The hope is to allow researchers to ask valuable questions using relevant models that are both biologically and clinically meaningful.

Features and constituents of a preclinical animal model that influence its translational relevance for development of cancer immunotherapy drugs can be classified in the following three segments: the tumor, the host, and the modality. Researchers looking for the most optimal preclinical model for their drug would want to pick something that has translational relevance in all three segments, but often, such perfect models may not exist and may require either de novo generation or compromising on certain attributes. Hence, there is no "one-size-fits-all" type of approach when it comes to optimizing the translational potential of preclinical animal models in immuno-oncology.

Tumor

A preclinical tumor model is not only expected to represent the tumor type and tissue pathology relevant to human indications, but also replicate the complex tumor microenvironment, particularly with respect to its immune-composition. While certain tumor models may replicate some of these features well, they may lack others making it a challenge to find the most representative model for preclinical research. With respect to tumor types, mouse models can be classified into four categories: (i) cell line-derived xenografts (CDX), (ii) patient-derived xenografts (PDX), (iii) syngeneic models (SM), and (iv) genetically engineered mouse models (GEMMs). While both CDX and PDX models are of human origin, they are usually implanted in immune-compromised mice lacking an adaptive immune system, the hallmark of an antitumor immune response, limiting their use to evaluation of anticancer agents and those that modulate the innate immune system to fight cancer. Due to the presence of a full complement of the immune system, the SM and GEMM systems which are derived from primary mouse tumors have recently gained popularity in preclinical discovery of cancer immunotherapies. There are about 50-60 well-known syngeneic cancer cell lines representing a variety of cancer types that can grow when allo-grafted in fully immune-competent syngeneic host mice. However, the repertoire of these tumor models seems to be low compared to the diverse panel of human cancers observed clinically as well as to the cell lines (CDX)/tissues (PDX) derived from them. Moreover, most syngeneic cancer cell lines originated from tumors induced by chemical mutagenesis in mice rather than formation of spontaneous neoplasms, questioning their relevance in such tumor indications. Despite this, a striking similarity between syngeneic cell lines and human cancer is their high immunogenicity and presence of spontaneous non-synonymous mutations, which is shown to correlate with antitumor immune responses in patients. For these reasons, syngeneic models are the workhorse for preclinical development of cancer immunotherapies and thus, the first chapter of this book will cover the baseline concepts of transplantable syngeneic models as well as their translational impact in immuno-oncology drug discovery.

To better mimic the spontaneous nature of disease origin and progression, there has been an increased interest in generating GEMM models as well as deriving cancerous cell lines from them, which can be transplanted into naïve mice. In contrast to syngeneic cell lines, the GEMM derived tumor cell lines harbor relatively fewer mutations, which limit their immunogenicity. Approaches aimed at preserving/enforcing the expression of immunogenic peptides or increasing their genetic diversity e.g. using X-ray, radiation or chemical mutagenesis may be beneficial for generating a diverse panel of syngeneic tumor models such as PDXs, that can be established in fully immune-competent mice for evaluating novel immune therapies. Whether these models would accurately represent the various clinical tumor types and their relevance in terms of the immune contexture observed in patients is yet to be determined.

Regarding tumor cell morphology and protein landscape, PDX models seem to be the best choice while the GEMM-derived models catch up, owing to their diversity and direct derivation from patient tumors. However, almost all tumor models except spontaneous GEMMs are established via subcutaneous implantation of cells or tissues under the skin, where they may not recreate the tissue complexity present in primary human tumors. Particularly, this is of concern in cancers of solid tissue. For instance, colorectal cancer (CRC) is characterized by formation of hyperplastic polyps along the lining of colon, a pathology that is not reproduced by subcutaneous implantation of human or mouse CRC cells. To further elaborate on this, Clementine Le Magnen in Chapter 3 focuses on the role of tissue microenvironment in cancer initiation and progression, as well as how this aspect is disregarded by the currently used transplantable tumor models in mice. Furthermore, subcutaneous tumor implantation in most cases, does not lead to formation of distant metastases commonly observed in patients with cancers of late-stage. Indeed, most oncology drugs entering the clinic are first tested on late stage patients who have...