Humanized mice have become a useful tool in the study of infectious disease, cancer, autoimmunity, and regenerative disease (Shultz et al. 2007). The two most commonly used humanized mice are the bone marrow–liver–thymus (BLT)‐humanized mouse and the CD34‐humanized mouse. BLT‐humanized mice are essentially chimeras containing a fully engrafted human immune system inside a mouse. They are made by surgical implantation of thymus and liver tissue underneath the renal capsule of severely immune‐compromised mice followed by intravenous injection of CD34+ hematopoietic stem cells (HSCs) (Figure 8.1). Depending on the strain of mouse, bone marrow ablation via irradiation or chemotherapeutics such as busulfan, may be required prior to stem cell injection. The BLT‐humanized mice demonstrate systemic repopulation with human T cells, B cells, monocytes, and macrophages (Melkus et al. 2006) and develop human thymic tissue, referred to as an “organoid” underneath the renal capsule (Figure 8.2). The thymic organoid produces all populations of developing T cells and is required for the development of adaptive immune responses. The CD34‐humanized mice are made more simply by ablating the bone marrow of severely immune‐compromised mice and injecting CD34+ HSC (Greiner et al. 1998). Another model with limited use due to rapid graft versus host disease (GvHD) is the peripheral blood mononuclear cell spelled out (PBMC)‐humanized mouse, which is made by injecting human PBMC into severely immune‐compromised mice (Ishikawa et al. 2005; Mosier et al. 1988). (The PBMC‐ and CD34‐humanized mouse models are addressed elsewhere in this book.)

Checkpoint inhibitors are a new class of therapeutics that are being used with increasing frequency in the treatment of cancers and their effectiveness has been demonstrated in clinical trials with either single or combination checkpoint inhibitor treatment (Barbee et al. 2015). This efficacy is achieved by enhancing the patient’s antitumor immune response and provides a new modality to fight cancer (Sharma and Allison 2015). At the time of this writing, the U.S. FDA has approved seven checkpoint inhibitors, all of which are monoclonal antibodies, including anti‐CTLA‐4 (ipilimumab), anti‐PDL1 (atezolizumab, avelumab, and durvalumab), and anti‐PD‐1 (nivolumab, pembrolizumab, and cemiplimab). Effectiveness of checkpoint inhibitors has been reported in several types of cancers including advanced melanoma, renal cell carcinoma, head and neck, bladder, and lung cancer previously treated unsuccessfully with standard therapies (Powles et al. 2014; Wolchok et al. 2013).

Despite these successes, serious adverse events have been reported in patients. These drugs have been associated with frequent immune‐related adverse events and are not recommended for patients with concomitant autoimmune disease (Abdel‐Wahab et al. 2016, 2018). Adverse events reported in patients include immune‐mediated pneumonitis, hepatitis, colitis, nephritis, and thyroid dysregulation (Abdel‐Wahab et al. 2017; Chow 2013; Kazandjian et al. 2016). Nivolumab was tested in nonhuman primates and some of the adverse events reported in patients were not identified during this nonclinical testing (Wang et al. 2014). We tested two checkpoint inhibitors in BLT immune humanized mice to determine if they could model these adverse events (Weaver et al. 2019).

Several immune humanized mouse models have been used in the study of checkpoint inhibitor therapy albeit not in the context of modeling immune‐related adverse events (Sanmamed et al. 2015; Vudattu et al. 2014; Waldron‐Lynch et al. 2012). The BLT‐humanized mice have been previously tested in our laboratory (Semple et al. 2019; Weaver et al. 2015; Yan et al. 2019a, 2019b) and our group was the first to evaluate the ability of BLT‐humanized mice to demonstrate adverse events associated with checkpoint inhibitor therapy (Weaver et al. 2019). Briefly, we showed in our nivolumab study that all dose levels (2.5–10 mg/kg) experienced immune‐mediated pathology in a wide range of organs, including lung, skin, liver, pancreas, adrenal gland, muscle, and bone. High‐dose‐treated mice showed mild myocarditis which has been observed in some patients (Heinzerling et al. 2016; Laubli et al. 2015). Our study of ipilimumab also showed broad immune‐mediated inflammation, but importantly showed colitis, which is a significant adverse event associated with this therapeutic that has been reported in patients (unpublished results). Both studies demonstrated the ability of this mouse model to show adverse events reported in humans and suggest its use as a tool to better understand the toxicities of checkpoint inhibitor therapies. Thus, this chapter will discuss critical considerations in using BLT‐humanized mice to study checkpoint inhibitor adverse events.

Following humanization procedures (described in detail in Yan et al. 2019a), mice are evaluated beginning eight weeks post‐surgery for humanization. We conduct monitoring over a minimum of three sampling intervals, conducted three to four weeks apart. At least two sequential bleeds, three to four weeks apart, are needed to show increasing human leukocyte numbers prior to use in the studies. An example of flow cytometry assessing human leukocyte subsets in PBMC are shown at approximately 12 weeks post‐surgery (Figure 8.3), demonstrating human and murine leukocytes in absolute numbers per microliter of blood collected. We also show the percentage of human CD45+ cells, T cells (CD3), B cells (CD20), and CD4 and CD8 populations of T cells. To understand the difficulties inherent in data analysis when using percentage humanization, Figure 8.4 shows alternate methods of gating the same sample of blood from an individual BLT‐humanized mouse. Panels 4a and 4c show the use of human CD45 only staining, with either a small or large population of gated events and show lower human percentages than if murine leukocytes are also considered. In panels 4b and 4d, both human and murine CD45 staining is shown and demonstrates much higher proportions of humanization whether using a small or large gate. While not shown in this figure, if the small gate is used, many human B cells are excluded from consideration. The overall range of humanization percentage for this sample varies from 28.5% to 52.9% even though the absolute cell number is unchanged, making it clear that use of percentage could yield misleading results. It should also be noted that while severely immune‐compromised mouse strains do not have a functional adaptive immune system, they do produce small numbers of lymphocytes as shown in panels 4b and 4d as gray events in the lymphocyte region. Therefore, if murine CD45 is not used in the determination of humanization, incorrect numbers or percentages might be used.