Immune checkpoint blockade (ICB) monotherapy shows early promise for the treatment of nasopharyngeal carcinoma (NPC) in patients. Nevertheless, limited representative NPC models hamper preclinical studies to evaluate the efficacy of novel ICB and combination regimens. In the present study, we engrafted NPC biopsies in non-obese diabetic-severe combined immunodeficiency interleukin-2 receptor gamma chain-null (NSG) mice and established humanized mouse NPC-patient-derived xenograft (NPC-PDX) model successfully. Epstein–Barr virus was detected in the NPC in both NSG and humanized mice as revealed by Epstein–Barr virus-encoded small RNA (EBER) in situ hybridization (ISH) and immunohistochemical (IHC) staining. In the NPC-bearing humanized mice, the percentage of tumor-infiltrating CD8+ cytotoxic T cells was lowered, and the T cells expressed higher levels of various inhibitory receptors, such as programmed cell death protein 1 (PD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) than those in blood. The mice were then treated with nivolumab and ipilimumab, and the anti-tumor efficacy of combination immunotherapy was examined. In line with paired clinical data, the NPC-PDX did not respond to the treatment in terms of tumor burden, whilst an immunomodulatory response was elicited in the humanized mice. From our results, human proinflammatory cytokines, such as interferon-gamma (IFN-γ) and interleukin-6 (IL-6) were significantly upregulated in plasma. After treatment, there was a decrease in CD4/CD8 ratio in the NPC-PDX, which also simulated the modulation of intratumoral CD4/CD8 profile from the corresponding donor. In addition, tumor-infiltrating T cells were re-activated and secreted more IFN-γ towards ex vivo stimulation, suggesting that other factors, including soluble mediators and metabolic milieu in tumor microenvironment may counteract the effect of ICB treatment and contribute to the tumor progression in the mice. Taken together, we have established and characterized a novel humanized mouse NPC-PDX model, which plausibly serves as a robust platform to test for the efficacy of immunotherapy and may predict clinical outcomes in NPC patients.

Background and Aims Recent development of multiple treatments for human hepatocellular carcinoma (HCC) has allowed for the selection of combination therapy to enhance the effectiveness of monotherapy. Optimal selection of therapies is based on both HCC and its microenvironment. Therefore, it is critical to develop and validate preclinical animal models for testing clinical therapeutic solutions. Approach and Results We established cell line–based or patient‐derived xenograft–based humanized‐immune‐system mouse models with subcutaneous and orthotopic HCC. Mice were injected with human‐specific antibodies (Abs) to deplete human immune cells. We analyzed the transcription profiles of HCC cells and human immune cells by using real‐time PCR and RNA sequencing. The protein level of HCC tumor cells/tissues or human immune cells was determined by using flow cytometry, western blotting, and immunohistochemistry. The HCC tumor size was measured after single, dual‐combination, and triple‐combination treatment using N‐(1ʹ,2‐Dihydroxy‐1,2ʹ‐binaphthalen‐4ʹ‐yl)‐4‐methoxybenzenesulfonamide (C188‐9), bevacizumab, and pembrolizumab. In this study, human immune cells in the tumor microenvironment were strongly selected and modulated by HCC, which promoted the activation of the IL‐6/Janus kinase 2 (JAK2)/signal transducer and activator of transcription 3 (STAT3) signaling pathway in tumor cells and led to augmented HCC proliferation and angiogenesis by releasing angiogenic cytokines in humanized‐immune‐system mice with HCC. In particular, intratumor human cluster of differentiation–positive (hCD14+) cells could produce IL‐33 through damage‐associated molecular pattern/Toll‐like receptor 4/activator protein 1, which up‐regulated IL‐6 in other intratumor immune cells and activated the JAK2/STAT3 pathway in HCC. Specific knockdown of the CD14 gene in human monocytes could impair IL‐33 production induced by cell lysates. Subsequently, we evaluated the in vivo anti‐HCC effect of C188‐9, bevacizumab, and pembrolizumab. The results showed that the anti‐HCC effect of triple‐combination therapy was superior to that of single or dual treatments. Conclusions Humanized‐immune‐system HCC mouse models are suitable for identifying targets from cancer and immune components and for testing combinational therapies.

In recent decades, chimeric antigen receptor (CAR)–engineered immune effector cells have demonstrated promising antileukemic activity. Nevertheless, their efficacy remains unsatisfactory on solid cancers, plausibly due to the influence of tumor microenvironments (TME). In a novel mouse cancer model with a humanized immune system, tumor-infiltrating immunosuppressive leukocytes and exhausted programmed death protein-1 (PD-1) high T cells were found, which better mimic patient TME, allowing the screening and assessment of immune therapeutics. Particularly, membrane-bound programmed death ligand 1 (PD-L1) level was elevated on a tumor cell surface, which serves as an attractive target for natural killer (NK) cell–mediated therapy. Hematopoietic stem cell–derived CAR-NK (CAR pNK) cells targeting the PD-L1 showed enhanced in vitro and in vivo anti-solid tumor function. The CAR pNK cells and nivolumab resulted in a synergistic anti-solid tumor response. Together, our study highlights a robust platform to develop and evaluate the antitumor efficacy and safety of previously unexplored therapeutic regimens.