Views: 0 Author: Site Editor Publish Time: 2026-11-11 Origin: Site
HKeyBio Industry Commentary
Humanized Mice or NHP Models: How Should You Choose?
HKeyBio Q&A 51 | Selecting Models Around the Evidence Needed
Originally published in Chinese on September 8, 2026. This commentary reflects information available as of September 8, 2026.
A1. Choose according to the evidence missing from the current decision, rather than a hierarchy of models.
Humanized mice primarily address efficacy mechanisms and patient heterogeneity in the context of human targets or human immune cells. Nonhuman primate (NHP) models primarily address systemic exposure, tissue distribution, systemic pharmacodynamics (PD), and safety in a pharmacologically relevant species. The two can complement one another, but they are not interchangeable in several critical settings.
A2. Prioritize a humanized mouse model matched to the mechanism, such as a hematopoietic stem cell (HSC) model, a peripheral blood mononuclear cell (PBMC) model, or a model incorporating human tissue or a human target. The key is not simply the presence of human cells: the target cells, tissue site, donor source, and method of disease induction must support the mechanism being measured. Differences among humanized models also matter.
Hu-PBMC models can be established relatively quickly and predominantly support mature peripheral immune responses, particularly T-cell responses. However, xenogeneic graft-versus-host disease (xeno-GVHD) can limit the experimental window, making these models unsuitable for some long-term efficacy or safety studies.
Hu-HSC models can generate T cells, B cells, and some myeloid cells. Human myeloid reconstitution, cytokine support, immune maturation, and human leukocyte antigen (HLA) restriction nevertheless depend on the host strain and humanization method. Hu-BLT (bone marrow–liver–thymus) models incorporate human liver and thymus tissue together with HSCs, providing a T-cell developmental environment with human HLA restriction. Their construction is more complex, with additional constraints involving tissue sourcing, reproducibility, and study timelines.
Figure 1. Construction of Hu-PBMC/PBL, Hu-HSC, and Hu-BLT mouse models. PBMC, peripheral blood mononuclear cell; PBL, peripheral blood lymphocyte. Reproduced from Chen et al. [1], Figure 2, under CC BY. © 2022 Chen et al.
A3. Assess NHP species relevance through binding, functional activity, expression profiles, cross-reactivity, pharmacokinetics (PK), and immunogenicity risk. Where needed, include receptor occupancy, tissue distribution, and the effects of anti-drug antibodies (ADAs) or immunogenicity on interpretation. NHP use is scientifically justified when the species has sufficient target and functional pharmacological relevance to the candidate and the study question requires intact systemic physiology.
A4. Neither can answer that question independently.
Safety evidence should integrate human tissue or cell studies in vitro, species relevance, in vivo models, exposure, and existing knowledge of the mechanism or therapeutic class. This is a scientific framework for combining evidence; it does not mean every program must follow the same animal study package. For biotechnology-derived pharmaceuticals, regulatory species selection and the nonclinical package should be considered in light of ICH S6(R1), the candidate’s attributes, and the development stage. [2]
Table 1. Humanized mice and NHPs: complementary strengths and evidence boundaries.
Dimension | Humanized mice | NHPs |
Human targets and human immune cells | Strength: direct investigation of human targets, donor differences, and selected human immune circuits, provided that humanization quality is adequate. | Generally cannot replace humanized mice for questions involving human donors, human HLA, or human-specific immune cell mechanisms. |
Systemic physiology and PK | Immune-reconstituted models retain an immunodeficient mouse host. Human cell maturity, tissue microenvironment, cytokine support, and lifespan remain limited. | Strength: a more intact context for systemic exposure, tissue distribution, and repeat dosing, provided the species is pharmacologically relevant. |
Efficacy and mechanism | Useful for early Go/No-Go decisions, mechanistic dissection, combination strategies, and validation of dependence on the human target. | Useful for systemic PD and additional translatable biomarkers in a relevant species; not inherently “closer to humans.” |
Safety | Can reveal selected human immune activation risks. Immune-reconstituted mice generally do not replace good laboratory practice (GLP) repeat-dose toxicology or a comprehensive safety assessment. | Can support systemic toxicology in a relevant species, while still potentially missing human-specific toxicities. |
Speed, samples, and reproducibility | Can run in parallel and include multiple donors. Control human chimerism, the GVHD window, donor batches, and microbiome background. | Longer timelines, higher costs, and greater ethical constraints; sample sizes are usually limited. Prioritize interpretable study questions and endpoints. |
A5. Five situations illustrate why model selection must follow the question and mechanism.
If the candidate binds the human target but not its NHP counterpart, NHPs generally cannot reliably evaluate efficacy, PD changes, or toxicity mediated by that target. A negative NHP result therefore does not establish the absence of pharmacological activity or on-target risk in humans. More appropriate evidence may come from human in vitro systems, mechanism-matched humanized mice or human-target mice, and surrogate molecules or alternative models supported by adequate pharmacological bridging.
The central objective is to integrate evidence showing whether the candidate has its intended pharmacological activity and potential risks in the human-target context.
Binding to the NHP target does not, by itself, establish that NHPs are a pharmacologically relevant species. Even when binding affinity is similar for the human and NHP targets, substantial differences in cellular functional activity, the relationship between receptor occupancy and effect, or downstream PD responses can alter interpretation of in vivo dose, efficacy, and the safety margin.
Binding is therefore the first step in assessing species relevance. It is also necessary to establish whether that binding produces functional pharmacology sufficiently comparable to that in humans. When functional relevance is still uncertain, a focused functional bridging assessment can clarify whether a larger NHP study is the most informative next step. [2]
Humanization does not mean that the entire human immune system has been reconstructed. If the candidate depends on human myeloid cells, tissue-resident immune cells, or a specific disease-associated tissue microenvironment, the selected model must adequately reproduce those components. High overall human immune cell engraftment alone does not justify interpreting an observed phenotype as evidence of the clinical mechanism.
First establish whether the cells, tissues, and signaling pathways that determine drug action are present and functional in the model. This is the foundation for a more reliable mechanistic interpretation of its PD changes.
Even a pharmacologically relevant NHP species may not identify every human-specific immune risk. Differences between humans and NHPs in target expression within key immune cell subsets, receptor distribution, or activation thresholds can limit the ability of an NHP pharmacological response to warn of human immune activation or cytokine release.
Such programs typically require integrated human in vitro evidence, including whole-blood or PBMC functional assays, cytokine release assays, and receptor occupancy and functional studies. Appropriate humanized mouse models can provide supplementary in vivo evidence where needed. These approaches address different levels of the question; no single model can independently establish human safety.
For long-term exposure, repeat dosing, and systemic safety risks across multiple organs, humanized mice generally have difficulty replacing systemic studies in a pharmacologically relevant species. Host physiology, immune system maturity, the tissue microenvironment, and the GVHD-limited window in some models constrain their ability to serve as the sole basis for long-term systemic safety assessment.
Conversely, NHPs cannot replace mechanistic studies centered on human donor differences, human HLA backgrounds, or human-specific targets. Long-term systemic safety and human-specific mechanisms often require evidence from different models, integrated into the overall preclinical assessment.
Figure 2. Donor-specific immune cell reconstitution kinetics in CD34+ HSC-engrafted NSG mice. Panels A–G show human CD45+ cells, B cells, total T cells, CD4+ and CD8+ T cells, NK cells, and CD56+CD3+ cells labeled “NK-T” in the original study. Reproduced from Rajagopal et al. [3], Figure 1, under CC BY. © 2026 Rajagopal, Ieong, Mate and Vessillier.
Assess sequence, binding, cellular function, tissue expression, and expected exposure in humans, mice, and cynomolgus or rhesus macaques, as applicable. Sequence homology or a single-point surface plasmon resonance (SPR) result alone does not define a relevant species.
Humanized mice address human mechanisms and efficacy hypotheses; NHPs address systemic questions in a relevant species; human in vitro systems add causal evidence at the tissue or cell level. Each model should have a clear primary question, primary endpoint, and interpretation of a negative or otherwise unsuccessful outcome. All three approaches are not necessarily required. The practical question is: which few critical pieces of evidence are still missing for the next Go/No-Go decision?
For example: “This study supports PD changes only in the context of human PBMC reconstitution,” or “This study does not support inference about long-term human safety.” Defining what the model can and cannot answer before a study begins is more important than relying on explanations of positive or negative results after it ends.
Model selection brings together a coherent evidence chain across five levels: target, cells, tissue, exposure, and risk. HKeyBio, the autoimmune disease model specialist, can combine humanized mouse, NHP, human in vitro functional, and PK/PD studies around the candidate’s species relevance, disease mechanism, and development stage. The aim is to design the minimum sufficient evidence package for the next decision. Our focus extends beyond which animal model to select to the specific question each model must answer for candidate development.
[1] Chen J, Liao S, Xiao Z, et al. The development and improvement of immunodeficient mice and humanized immune system mouse models. Front Immunol. 2022;13:1007579. Source
[2] ICH. S6(R1): Preclinical Safety Evaluation of Biotechnology-Derived Pharmaceuticals. 2011. Source
[3] Rajagopal D, Ieong KS, Mate R, Vessillier S. Evaluation of a reference antibody panel for prediction of cytokine release in humanised mouse models in vivo. Front Immunol. 2026;17:1736130. Source
Author: HKeyBio Marketing Department | Scientific review: Jing Yan | Editor: Wang Shuyan
HKeyBio, the autoimmune disease model specialist, is a preclinical contract research organization (CRO) focused on autoimmune and allergic diseases, providing end-to-end in vivo efficacy services. Its core technical team has more than 20 years of experience across the autoimmune drug development process. The company is ISO 9001 certified and operates molecular, cellular, pathology, small-animal, and analytical testing facilities in Suzhou Industrial Park, together with NHP research sites in Suzhou and Guangxi. It has nearly 80 employees, with project leaders holding master’s or doctoral degrees. The team has accumulated experience supporting more than 500 investigational new drug (IND) applications in autoimmune diseases and helps address complex challenges across drug development and regulatory submission.
· 50+ NHP models of autoimmune and allergic diseases
· 500+ animal models of autoimmune and allergic diseases
· Experience supporting 500+ IND applications in autoimmune diseases
