Humanized Immune System Models: Bringing In Vivo Immune Context to Translational Immuno-oncology

Why In Vivo Immune Context Matters

Immuno-oncology therapies are often developed around defined mechanisms: target engagement, immune-cell activation, checkpoint modulation, cytokine release, tumor-cell killing, or immune-mediated remodeling of the tumor microenvironment. These mechanisms can be studied with precision in controlled experimental systems, and those studies remain essential for determining whether a therapeutic concept is biologically plausible.

But many immuno-oncology therapies do not act through isolated molecular events alone. Their activity may depend on whether immune cells traffic into tumors, persist after activation, maintain effector function, interact with suppressive tumor microenvironments, or respond to checkpoint modulation under exposure conditions. These processes involve spatial, systemic, and multi-compartment biology that cannot be fully reproduced in reductionist systems.

This creates a central challenge for translational immuno-oncology: the model must be aligned not only with the molecule being tested, but with the biology the therapy is intended to engage. A model that captures target binding or cytotoxicity may not capture immune suppression, tumor infiltration, myeloid involvement, cytokine biology, or adaptive resistance. Mechanistic evidence is therefore necessary, but not always sufficient.

Humanized immune system (HIS) mouse models help address this gap by introducing human immune components into an in vivo setting. Within the limitations of each humanization strategy, HIS models allow researchers to study human immune-tumor interactions in living systems where tumor growth, immune infiltration, exposure, tissue distribution, and systemic immune responses can be evaluated together.

The question is not whether in vitro or in vivo systems are better. The more useful question is where mechanistic evidence ends and where in vivo immune context becomes necessary.

When Immune Context Changes Translational Interpretation

The translational importance of HIS models becomes clearest when the presence or absence of human immune biology changes the interpretation of therapeutic response.

A collaboration between the University of Leuven and TransCure bioServices provides a direct example. In a study of pS6high uterine leiomyosarcoma, De Wispelaere and colleagues evaluated PI3K/mTOR pathway inhibition and PD-1 blockade using CD34+ HSC-humanized patient-derived xenograft models to assess immune-dependent mechanisms of response and resistance. The key translational point is that immunodeficient and humanized model readouts were not equivalent: in nude mice, PI3K/mTOR inhibition produced an initial tumor response, but tumors rapidly developed resistance and resumed growth; in CD34+ humanized mice, the same pathway inhibition produced more prolonged tumor growth suppression, supporting the interpretation that the reconstituted immune system contributed to tumor control.

This distinction matters because the direct tumor-cell effect of a therapy may represent only part of the mechanism. In this study, PI3K/mTOR inhibition was associated with increased CD4+ and CD8+ T-cell infiltration, macrophage repolarization toward an anti-tumorigenic phenotype, enhanced dendritic-cell antigen presentation, and upregulation of antigen-processing and presentation pathways in tumor cells. When PI3K/mTOR inhibition was combined with anti-PD-1 therapy, the study reported partial or complete tumor responses, while single-agent anti-PD-1 did not produce tumor growth control.

The study therefore illustrates a central risk in translational model selection: an immunodeficient model may capture the tumor-autonomous component of response while missing immune-dependent efficacy. In this case, the immune-dependent component was not a minor refinement to the readout. It was central to interpreting whether the therapeutic strategy could produce durable tumor control and whether PD-1 blockade was meaningfully evaluable in combination.

The same principle extends beyond a single tumor type or pathway: when human immune reconstitution changes the therapeutic readout, the model is no longer just a permissive host — it becomes part of the translational interpretation.

A metastatic prostate cancer study from the Kregel lab reinforces the same principle from another angle. Investigators used humanized NOG and humanized NOG-EXL mice engrafted with human CD34+ hematopoietic stem cells to model prostate cancer in the presence of human immune components. The significance of that work was not simply that tumors could be grown in humanized mice, but that immune context altered biological interpretation. Therapeutic response to androgen receptor inhibition more closely aligned with clinical outcomes only when human immune cells were present, and some therapeutic effects were minimal or absent in immunodeficient hosts.

Together, these examples reinforce a central principle for translational immuno-oncology: immune context can change what a model teaches you. In some settings, the immune compartment is not merely an experimental addition. It is part of the biology required to determine whether a therapeutic mechanism remains meaningful in vivo.

Matching Humanized Platforms to Immune-Engaging Therapeutic Mechanisms

The same model-selection logic appears in therapeutic development programs where the mechanism depends on human immune-cell behavior. The relevant model is not always the same type of humanized mouse. For some mechanisms, the priority is multilineage immune reconstitution; for others, the key requirement is durable support of a defined human effector-cell population.

In a study of SAR445514, a trifunctional NK-cell engager targeting BCMA in multiple myeloma, Tang and colleagues evaluated a molecule designed to co-engage NKp46 and FcγRIIIa on NK cells while targeting BCMA on myeloma cells. Because this mechanism depends specifically on human NK-cell activity, the in vivo studies used a specialized human NK-cell-supported hIL-15 NOG approach rather than a generic HIS platform. Expanded human NK cells were adoptively transferred into irradiated hIL-15 NOG mice before MM1.R tumor engraftment, enabling dose-dependent evaluation of anti-myeloma activity in vivo.

T-cell engagers raise a related but distinct model-selection question. Obermajer and colleagues described JNJ-78306358, a bispecific T-cell engaging antibody targeting CD3 and HLA-G. The study reports that JNJ-78306358 induced T cell-mediated cytotoxicity of HLA-G-expressing tumors in vitro and in vivo, and that the preclinical activity supported clinical evaluation. Its in vivo studies included T cell-humanized NSG mice and CD34+ HSC-humanized models using NOG-EXL or NSG-SGM3 hosts. These humanized systems contributed to development by enabling evaluation of anti-tumor activity, T-cell infiltration, and activity across human cell line- and patient-derived xenograft models in settings where human T-cell engagement was central to the therapeutic mechanism.

Together, these examples show how humanized in vivo systems can be selected around the immune mechanism being tested. They do not suggest that one model type fits every program. Instead, they reinforce the core point of this article: when therapeutic activity depends on human immune-cell engagement, persistence, infiltration, or checkpoint biology, model selection can determine whether the resulting data are interpretable.

Designing HIS Studies Around the Biology

Humanized model selection should begin with the biological question, not with a generic preference for one platform. Different HIS approaches support different study windows, immune compartments, and interpretive strengths. Selecting the right model requires understanding the therapeutic mechanism, tumor context, immune compartment of interest, and readout strategy. These models should not be interpreted as complete replicas of human immunity; their value depends on matching the humanization strategy and host background to the immune biology being tested.

A PBMC-based model may be appropriate for certain short-term studies focused on mature human T-cell activity. A CD34+ HSC-engrafted model may be more appropriate when broader immune development, longer study duration, or multilineage immune context is needed. Myeloid-supportive models such as huNOG-EXL can be useful when myeloid biology, tumor immune microenvironment features, or immune-suppressive mechanisms are relevant, while also requiring careful study planning because cytokine-supported humanization can introduce model-specific health and timing considerations.

Model choice should also account for tumor placement, tumor type, therapeutic modality, and endpoint design. Orthotopic tumor engraftment may be important when anatomical context is relevant. Patient-derived xenografts may be valuable when tumor architecture or patient-derived biology is central to the question. Flow cytometry, histopathology, cytokine analysis, imaging, and tumor rechallenge studies may each answer different aspects of the translational question.

For immuno-oncology programs, useful HIS study design often depends on a small number of linked decisions: which immune compartment is required, whether the tumor model preserves the relevant biology, whether the study window fits the model, and whether the readouts can detect the immune response being tested. The more closely these design decisions align with the therapeutic biology, the more informative the resulting data are likely to be.

Putting HIS Model Selection Into Practice

Translating HIS model potential into useful preclinical evidence requires more than access to a humanized mouse model. The tumor model, humanization strategy, study timeline, immune readouts, and endpoint plan must all align with the therapeutic mechanism.

For teams translating these questions into executable studies, model selection must be connected to tumor model strategy, immune profiling, endpoint selection, and in vivo pharmacology execution. TransCure bioServices supports this step through humanized in vivo oncology study design and execution, helping align the model system with the immune biology and translational question under investigation.

That role is most valuable when it remains anchored in the science. The objective is not simply to run a humanized mouse study. The objective is to select and execute a study design that can answer the biological question the program depends on.

Conclusion: Translational Insight Requires Immune Context

Mechanistic assays remain essential for understanding how an immuno-oncology therapy is intended to work. They can define target engagement, immune-cell activation, tumor-cell killing, and early pathway effects with a level of control that is difficult to achieve in vivo.

But for therapies that depend on human immune-cell behavior, tumor infiltration, cytokine signaling, checkpoint modulation, myeloid biology, or multi-compartment immune regulation, mechanism alone is not enough. The translational question is whether that mechanism remains meaningful in a living system where human immune components interact with tumors under biologically relevant conditions.

Humanized immune system mouse models provide an in vivo framework for evaluating these questions. They do not fully reproduce human immunity, and their interpretation must account for the specific humanization strategy, host strain, donor source, tumor model, and study design. Used appropriately, however, they can reveal immune-dependent biology that may be absent, muted, or uninterpretable in conventional immunodeficient models.

The studies highlighted here point to a common principle: immune context can change translational interpretation. In the University of Leuven and TransCure bioServices PI3K/mTOR study, an immunodeficient model captured only a transient tumor response, while the humanized model revealed immune-dependent tumor control and the evaluability of PD-1 blockade in combination. Other studies using NK-cell-engrafted, T cell-humanized, CD34+ HSC-humanized, or humanized patient-derived model systems reinforce the broader point: when therapeutic activity depends on human immune biology, model selection can determine whether the resulting data are interpretable.

For immuno-oncology researchers, the practical implication is straightforward: the model should be chosen for the immune biology the program needs to interpret. By aligning humanized model choice, tumor model strategy, immune readouts, and study execution, researchers can generate evidence that is more informative for translational decision-making.

References

De Wispelaere W, et al. PI3K/mTOR inhibition induces tumour microenvironment remodelling and sensitises pS6high uterine leiomyosarcoma to PD-1 blockade. Clinical and Translational Medicine. 2024 May 6;14(5):e1655. doi: 10.1002/ctm2.1655

Kostlan RJ, et al. Clinically Relevant Humanized Mouse Models of Metastatic Prostate Cancer Facilitate Therapeutic Evaluation. Mol Cancer Res. 2024 Sep 4;22(9):826-839. doi: 10.1158/1541-7786.MCR-23-0904.

Tang A, et al. Targeting BCMA in multiple myeloma with a trifunctional NK cell engager. Cell Reports Medicine. 2026 Mar 17;7(3):102628. doi: 10.1016/j.xcrm.2026.102628.

Obermajer N, et al. JNJ-78306358, a first-in-class bispecific T cell engaging antibody targeting CD3 and HLA-G. iScience. 2025 Feb 4;28(3):111876. doi: 10.1016/j.isci.2025.111876.