Challenges in the nonclinical safety evaluation of biotechnology-derived pharmaceuticals and perspectives for revision of the ICH S6 guideline based on analysis of approved products in Japan

Abstract

This article outlines current challenges in the nonclinical safety evaluation of biotechnology-derived pharmaceuticals and discusses future directions for revising the ICH S6 guideline. Analysis of products approved in Japan showed that, despite the frequent use of nonhuman primates for antibody therapeutics, novel and unexpected toxicities are rarely detected in long-term toxicity studies. We also summarize the potential utility and challenges of using homologous proteins, transgenic animals, and new approach methodologies. Based on these findings, we propose that the nonclinical safety evaluation process should transition to a flexible, scientifically sound framework based on the weight of evidence, rather than regarding standardized animal testing as the default approach.

Highlights

This article outlines key issues in the nonclinical safety evaluation of biopharmaceuticals. It presents perspectives to inform future revision of the ICH S6 guideline. These perspectives are based on the use of nonhuman primates, the role of long-term toxicity studies, and the potential application of new approach methodologies.

Introduction

The ICH S6 guideline [1], issued by the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) in 1997, continues to serve as the fundamental guidance for the nonclinical safety evaluation of biotechnology-derived pharmaceuticals (biopharmaceuticals). This guideline is based on the principle that the toxicity of biopharmaceuticals should be evaluated on a case-by-case basis, considering their characteristics. This approach reflects that these products are manufactured using diverse biologically derived expression systems.

Recently, as modalities in biopharmaceutical development have become more diverse, cases have emerged in which conventional toxicity studies using traditional laboratory animals cannot adequately evaluate safety. In addition, nonhuman primates (NHPs) have been widely used as relevant species in the toxicity evaluation of biopharmaceuticals; however, practical constraints, such as limited availability and rising costs, have become obstacles to drug development. Given this situation, both biopharmaceutical developers and regulatory authorities are increasingly emphasizing the need to clarify approaches to reduce NHP use, minimize studies, shorten study duration, reevaluate species selection principles, and clarify the role of alternative evaluation methods when an appropriate animal species is unavailable.

In April 2025, the U.S. Food and Drug Administration (FDA) Science Board presented comprehensive recommendations for the regulatory implementation of new approach methodologies (NAMs) that do not use animals. These recommendations considered the scientific limitations of animal studies [2]. The FDA then published a roadmap describing a phased approach to reducing animal testing for monoclonal antibodies based on NAMs [3, 4]. These initiatives aim not only to address the 3Rs in animal experimentation but also to improve the predictability of toxicity by aligning safety evaluations with human biological characteristics.

We have investigated the issues biopharmaceutical developers and regulatory authorities face when implementing the ICH S6 guideline, with the aim of proposing recommendations that are feasible from a regulatory perspective. In this review article, recognizing that the regulatory implementation of NAMs and the shift toward reducing animal testing outlined in the FDA roadmap could be important factors in the future revision of the ICH S6 guideline, we summarize our analysis of the challenges and future perspectives in the nonclinical safety evaluation of biopharmaceuticals.

Challenges Related to the Use of NHPs

The ICH S6 guideline recommends using relevant species, that is, species in which the test material is pharmacologically active, for repeated-dose toxicity studies and reproductive and developmental toxicity studies of biopharmaceuticals. Based on Pharmaceuticals and Medical Devices Agency (PMDA) review reports [5], we investigated the use of NHPs in repeated-dose toxicity studies for 134 biopharmaceutical products approved in Japan between January 2012 and March 2024. Overall, NHP use was confirmed for 119 of the 134 biopharmaceuticals and for 77 of the 84 antibody therapeutics (including IgG antibodies, bispecific antibodies, antibody fragments, and antibody–drug conjugates) (Fig. 1). This demonstrates that the nonclinical safety evaluation of biopharmaceuticals in Japan remains highly dependent on NHPs. At the same time, NHP use is associated not only with practical issues such as limited animal resources and ethical considerations, but also with scientific limitations. For example, human safety and toxicity cannot necessarily be adequately predicted. For instance, the anti-CD28 super-agonist antibody TGN1412 caused a severe cytokine storm during a first-in-human study. However, this outcome was not predicted in cynomolgus monkeys used for toxicity studies because the distribution of the target between monkeys and humans differed [6]. In addition, immune-related adverse events (irAEs) affecting the skin, gastrointestinal tract, liver, and other organs have been observed clinically with immune checkpoint inhibitors, including anti-CTLA-4, anti-PD-1, and anti-PD-L1 antibodies. These characteristic irAEs have not necessarily been reproduced in nonclinical toxicity studies, including those in NHPs [7]. Thus, conventional NHP studies alone may be insufficient for toxicity prediction in some cases.

Fig. 1.

Nonhuman primates (NHPs) in repeated-dose toxicity studies for biopharmaceuticals approved in Japan (January 2012–March 2024)

The ICH S6 guideline requires repeated-dose toxicity studies of at least 6 months to support a marketing application for biopharmaceuticals intended for long-term administration for chronic diseases. However, it has recently been debated whether long-term repeated-dose toxicity studies in NHPs provide additional value for the safety evaluation of biopharmaceuticals, particularly antibody therapeutics [8, 9].

Accordingly, among the 77 antibody therapeutics approved in Japan for which studies in NHPs were confirmed, we compared toxicity profiles between short-term and long-term NHP studies (Fig. 2). For this analysis, repeated-dose toxicity studies of up to 3 months (77 products) were defined as short-term NHP studies, and repeated-dose toxicity studies of 6 months or longer (58 products) were defined as long-term NHP studies. In Fig. 2, injection-site reactions and the effects of neutralizing antibodies were excluded from the toxicity findings. Repeated-dose toxicity studies of 9 months or longer in NHPs had been conducted for 12 products.

Fig. 2.

Comparison of toxicological findings between short-term and long-term nonhuman primate (NHP) studies for antibody therapeutics approved in Japan. The numbers shown for each category in panel B indicate the number of products with findings. When a product corresponded to multiple categories, it was counted in each category; within a given category, each product was counted only once.

The results showed that in both short- and long-term NHP studies, more than 90% of products showed no noteworthy toxicological findings or only pharmacology-related findings. Toxicity targets were generally consistent across short-term and long-term studies. By contrast, in long-term studies, increases in the severity or extent of pharmacology-related toxicity, or decreases in the no-observed-adverse-effect level (NOAEL), were observed for approximately 20% of products. The proportion of studies showing toxicities with unclear relationships to pharmacological activity was slightly higher in long-term studies (5.2%) than in short-term studies (2.6%). Among these, renal glomerular injury observed with an activin receptor IIB/Fc fusion protein had an unclear mechanism, was observed in both short- and long-term NHP studies, and was also reported in clinical trials. Accordingly, information on renal dysfunction was provided in the package insert in some cases [10]. Furthermore, Table 1 shows toxicological findings observed in long-term NHP studies but not in short-term studies. Many of these findings appear to reflect either pharmacological effects that became apparent during long-term administration or worsening effects in target tissues and organs. Although many toxicological findings newly detected in long-term studies appear to be predictable from similar products or their pharmacological activity, when information on the target molecule or related drugs is insufficient, such findings may warrant attention in the planning and management of clinical trials.

Table 1.Toxicological findings not observed in short-term nonhuman primate (NHP) studies but detected in long-term studies of antibody based therapeutics

Therapeutic modality and target	Findings first observed in long-term NHP studies
Pharmacology-related findings
Fusion protein composed of VEGFR/Fc	Ulceration of respiratory epithelium
mAb to VEGFR-2	Proteinuria*, Glomerulonephritis*, Epiphyseal dysplasia*
mAb to EGFR	Dermatitis
Trivalent bispecific Ab to TNFα/albumin	Inflammation* (infection)
mAb to IL-17A/IL-17F	Infection, Inflammation
mAb to IL-17R	Infection, Inflammation
mAb to PD-L1	Menstrual cycle irregularities
Off-target or mechanism-uncertain toxicity
Fusion protein composed of activin receptor type IIB/Fc	Degeneration of renal tubules, hemorrhage in kidney
mAb to type I interferon receptor	Arteritis*
Antibody fragment to neonatal FcR	Lymphadenopathy

*; Findings observed in repeated dose toxicity studies of 9 months or longer, conducted as long-term studies.

Although these results are based on a limited survey of products approved in Japan, toxicological profiles are consistent with previous studies in showing that most toxicological findings observed in both short- and long-term NHP studies of antibody therapeutics are predictable and related to the pharmacological action of the antibody. At the same time, a small number of toxicological findings with an ambiguous relationship to pharmacological activity were also identified. These findings should be evaluated in terms of frequency, type, characteristics, and the impact of toxicity on clinical trial planning and post-marketing safety evaluation.

For antibody therapeutics, short-term studies often provide sufficient information on the toxicity profiles relevant to adverse events in clinical trials. By contrast, long-term NHP studies may contribute to clinical trial planning and post-marketing safety evaluations. Therefore, the decision to omit long-term NHP studies should be made on a case-by-case basis. Such a judgment should not be based on a single factor, but rather on an integrated evaluation of information related to the mechanism of action, together with relevant information from nonclinical and clinical studies, including related products. For conventional monospecific IgG antibodies, evidence regarding the significance of long-term NHP studies is gradually accumulating. On the basis of this evidence, it may be easier to implement flexible evaluations, including the omission of long-term studies. By contrast, careful discussion is required even for monospecific IgG antibodies when knowledge of the mechanism of action is insufficient or when relevant toxicities or adverse events have been observed in short-term NHP studies or clinical trials.

Although Table 2 is not based on specific case studies, it summarizes our current views on situations in which the significance of conducting long-term studies is considered greater. In addition, the ICH S6 guideline recommends a carcinogenicity assessment based on various sources of information, including data from long-term repeated-dose toxicity studies. Therefore, how carcinogenic risk is assessed and positioned within the overall evaluation may also be an important factor in determining the necessity of long-term studies.

Table 2.Antibody therapeutics for which the need for long-term studies may be greater

Situations in which the need for long-term studies may be greater
First-in-class candidates (when knowledge of the mechanism of action is limited)
Candidates conjugated with chemically synthesized substances (e.g., small molecules or linkers)
Cases in which toxicological findings with an unclear mechanism are observed in studies of up to 3 months
Cases in which the target molecule is widely distributed throughout the body and diverse toxic effects may occur
Cases in which toxicities that are difficult to manage in clinical trials are observed and NOAEL information from long-term studies is considered useful
Cases in which special routes of administration (e.g., inhalation or intravitreal administration) indicate a need to evaluate local effects on tissues or organs in NHPs
Cases in which long-term studies are considered useful for carcinogenicity risk assessment
Cases in which sustained exposure is associated with toxicity raising concerns regarding reversibility (e.g., recycling antibodies)
Cases in which the issues to be examined in long-term studies and the significance of conducting such studies are clear (e.g., biomarker exploration)

NOAEL: no-observed-adverse-effect level; NHPs: Nonhuman primates.

In December 2025, the U.S. FDA issued draft guidance on antibody therapeutics [11], stating that for monospecific antibodies, long-term repeated-dose toxicity studies in non-rodents beyond 3 months are unnecessary when a weight of evidence (WoE)-based risk assessment sufficiently supplements data from a 3-month repeated-dose toxicity study in non-rodents as part of the nonclinical safety information. A future revision of the ICH S6 guideline should enable more flexible, scientifically sound decision-making considering this perspective. In addition, further clarification will be needed regarding the range of biopharmaceuticals for which long-term studies may be omitted, as well as the kinds of decisions that should be made under specific circumstances.

Challenges in Evaluation Using Homologous Proteins and Transgenic Animals Expressing the Human Target Molecule

When no relevant species exists for a clinical candidate, the ICH S6 guideline identifies proteins homologous to the clinical candidate (homologous proteins) and transgenic animals that express the human target molecule (human target-expressing transgenic animals) as options for toxicity studies. Of the 134 biopharmaceuticals approved in Japan, homologous proteins were used in eight products for repeated-dose toxicity studies and eight products for reproductive and developmental toxicity studies. By contrast, the use of human target-expressing transgenic animals was limited to one product in repeated-dose toxicity studies and two products in reproductive and developmental toxicity studies (Table 3).

Table 3.Examples of the use of homologous proteins and human target-expressing transgenic animals

INN	Year	Modality	Molecular description	Study / model used*
Certolizumab pegol	2012	Antibody fragment	PEGylated humanized anti-TNF antibody fragment (Fab)	Developmental and reproductive toxicity study / mouse homologous antibody (2, GLP)
Alemtuzumab	2014	IgG-type antibody	Humanized anti-CD52 antibody	Developmental and reproductive toxicity study / human CD52 transgenic mouse (2, GLP)
Secukinumab	2014	IgG-type antibody	Human anti-IL-17A antibody	Developmental and reproductive toxicity study / mouse homologous antibody (3, GLP)
Mepolizumab	2016	IgG-type antibody	Humanized anti-IL-5 antibody	Developmental and reproductive toxicity study / rodent homologous antibody (2, GLP)
				Other study / mouse homologous antibody (7 weeks, 2, non-GLP)
Pembrolizumab	2016	IgG-type antibody	Humanized anti-PD-1 antibody	Other study / mouse homologous antibody (29 days, 1, non-GLP)
Sarilumab	2017	IgG-type antibody	Human anti-IL-6 receptor α subunit antibody	Developmental and reproductive toxicity study / mouse homologous antibody (3, GLP)
				Juvenile animal study / mouse homologous antibody (9 weeks, 3, GLP)
Daratumumab	2017	IgG-type antibody	Human anti-CD38 antibody	Repeated-dose toxicity study / monkey homologous antibody (2 weeks, 2, non-GLP)
Dupilumab	2018	IgG-type antibody	Human anti-IL-4Rα subunit antibody	Repeated-dose toxicity study / mouse homologous antibody (5 weeks, 4, non-GLP)
				Repeated-dose toxicity study / monkey homologous antibody (~6 months, 2–4, GLP)
				Developmental and reproductive toxicity study / mouse homologous antibody (3, GLP)
				Developmental and reproductive toxicity study / monkey homologous antibody (2, GLP)
Blinatumomab	2018	Bispecific antibody	Mouse anti-CD19 / anti-CD3 bispecific antibody	Repeated-dose toxicity study / mouse homologous antibody (~13 weeks, 2–3, GLP)
				Developmental and reproductive toxicity study / mouse homologous antibody (2, GLP)
Ravulizumab**	2019	IgG-type antibody	Humanized anti-human complement C5 antibody	Repeated-dose toxicity study / mouse homologous antibody (~26 weeks, 2–3, GLP)
				Developmental and reproductive toxicity study / mouse homologous antibody (2, GLP)
Burosumab	2019	IgG-type antibody	Human anti-human FGF23 antibody	Other study / mouse homologous antibody (2 weeks, 4, non-GLP)
Inebilizumab**	2021	IgG-type antibody	Humanized anti-CD19 antibody	Repeated-dose toxicity study / human CD19 transgenic mouse (~26 weeks, 2–3, GLP)
				Developmental and reproductive toxicity study / human CD19 transgenic mouse (2, GLP)
Pabinafusp alfa	2021	Fusion protein	Humanized anti-transferrin receptor antibody fused to iduronate-2-sulfatase	Repeated-dose toxicity study / homologous antibody with equivalent affinity
				for humans and cynomolgus monkeys (4 weeks, 1, non-GLP)
Polatuzumab vedotin	2021	Antibody-drug conjugate	MMAE-conjugated humanized anti-CD79b antibody	Repeated-dose toxicity study / monkey homologous antibody (10 weeks, 2, GLP)
Vestronidase alfa	2022	Enzyme	β-Glucuronidase	Repeated-dose toxicity study / mouse homologous protein (13 weeks, 3, non-GLP)
Spesolimab**	2022	IgG-type antibody	Humanized anti-human IL-36 receptor antibody	Repeated-dose toxicity study / mouse homologous antibody (26 weeks, 2, GLP)
				Developmental and reproductive toxicity study / mouse homologous antibody (2, GLP)
Lecanemab	2023	IgG-type antibody	Humanized anti-human soluble amyloid-β aggregate antibody	Other study / human Aβ transgenic mouse and mouse homologous antibody
				(~18 weeks, 1–4, non-GLP)
Luspatercept	2024	Fusion protein	Modified human activin receptor type IIB–Fc fusion protein	Other study / mouse homologous model (13 weeks, 1, non-GLP)

*; Parenthetical information indicates study duration for repeat-dose toxicity studies, number of treated groups, and good laboratory practice (GLP) compliance, **; nonhuman primates (NHPs) were not used in repeat-dose toxicity studies.

Under the current ICH S6 guideline, evaluation using homologous proteins and transgenic models is positioned primarily for hazard identification related to pharmacological activity. It has been difficult to conduct a comprehensive risk assessment based on comparisons of exposure between humans and animals. In addition, the use of these models appears to be limited by practical constraints, including the time and cost required to generate them and the difficulty of interpreting results when the test material differs from the antibody under development.

By contrast, from the perspective of reducing NHP use, the potential use of homologous proteins and transgenic animals has recently attracted renewed attention. For antibody–drug conjugates, in which antibodies are chemically conjugated to small molecules, and for bioconjugates in which antibodies are linked to proteins, human-target-expressing transgenic animals may help clarify toxicity findings when species differences exist between humans and the test animal species in the target molecule recognized by the antibody moiety. In the FDA roadmap, such animals are listed as one of the transitional measures to reduce dependence on conventional animal studies [4]. In addition, the European Medicines Agency (EMA) states in its 2025 Reflection Paper [12] that surrogate molecules are useful for hazard identification and recommends using them to reduce or replace NHP studies. The EMA also notes that transgenic animals may be considered an alternative in vivo approach to NHP studies.

We broadly agree with these trends and basic concepts. However, to use homologous proteins and transgenic animals more appropriately in the future, we consider that further clarification is needed regarding the situations in which their use is envisioned, the applicable study conditions, the limitations of evaluation, and how to address these limitations. They may be particularly useful in the following situations: (1) when pharmacological responses in NHPs are insufficient compared with those in humans; (2) use as part of a WoE evaluation intended to justify the omission of long-term NHP studies; (3) to elucidate the mechanisms underlying toxicity; and (4) when the pharmacological target molecule does not exist in the animal species.

In such situations, it is important to avoid unnecessarily constraining study designs within the conventional framework of repeated-dose toxicity studies. Instead, study designs should be established according to the role and purpose of the study. In other words, group composition (including the need to evaluate sex differences), the number of animals per group, study duration (including dosing and recovery periods), and study endpoints should be designed to suit objectives such as hazard identification and mechanistic elucidation. In a future revision of the ICH S6 guideline, the document should not limit the use of these animal models to predefined cases but rather present a framework for developing a scientifically sound study design based on the nature of the biopharmaceuticals, the characteristics of the animal models, and the purpose of the study.

It is also important to clarify what information should be provided regarding the quality attributes of the test material and target molecules when homologous proteins or transgenic animals are used. Furthermore, because good laboratory practice (GLP) compliance may be difficult for studies using transgenic models, it will be necessary to reconsider and clarify how study data obtained in vitro or in vivo that are not necessarily GLP-compliant should be positioned to support marketing authorization. This issue is closely related to the discussion of NAMs described below.

Clarifying the ICH S6 guideline with these points in mind is expected to promote the practical use of homologous proteins and transgenic animals, reduce competition for limited NHP resources, and improve efficiency and sustainability of drug development. At the same time, if evaluations shift from NHPs to homologous proteins or transgenic animals using rodents, the total number of animals used could increase. Therefore, it is crucial to consider consistency carefully with the 3R principles. Moving forward, it will also be important to seek international consensus on balancing animal welfare with scientific and developmental rationality.

Potential Utility of Non-animal Evaluation Methods, including NAMs

Evaluation methods in vitro, using cells and specimens derived from animals and humans, have played a complementary role in nonclinical safety evaluations. They have been useful for assessing toxicities that are difficult to evaluate adequately in animal studies and for interpreting human-specific physiological and molecular biological characteristics, as well as interspecies differences between humans and animals. Advances in technologies such as organoids and culture systems that mimic microenvironments in vivo, including microphysiological systems (MPS), have increasingly made available evaluation systems that can reproduce certain functions and responses that resemble more closely those of living organisms [13, 14]. Against this background of technological progress, NAMs, as a collective term for such novel evaluation methods, are positioned as tools expected to expand further within the conventional nonclinical evaluation framework. However, from a scientific perspective, the complete replacement of animal studies by NAMs alone remains difficult in many respects [15]. For the integrated evaluation of biological responses, including whole-body exposure, interactions among multiple organs, and effects on the immune system, pregnancy, and growth, animal studies remain an important source of information. Therefore, at present, it is considered realistic to use NAMs not as a simple substitute for animal studies, but as part of an integrated evaluation combined with other nonclinical data. In other words, NAMs should be used as one component of a WoE approach. For antibody therapeutics targeting the same indication and target molecule, existing nonclinical and clinical data may, for example, enable extrapolation to potential target tissues or organ systems and associated toxicological findings. In such cases, it is scientifically reasonable to position similar products as comparators and to use systems in vitro based on human cells or tissues derived from toxicity target organs to conduct toxicity evaluations that closely correlate with human adverse events.

This perspective on the use of NAMs should be more clearly reflected in the next revision of the ICH S6 guideline. Currently, the guideline indicates that, when neither transgenic animal models nor homologous proteins are available, certain aspects of toxicity can be evaluated in a limited toxicity study in a single species. Such a study may take the form of a repeated-dose toxicity study lasting up to 14 days, including assessments of important functional endpoints, such as the cardiovascular and respiratory systems. This approach is considered to reflect the ethical view that safety should be confirmed in animals before clinical trials. However, given that studies in animal species lacking the pharmacological target molecule do not provide scientifically appropriate toxicity evaluations, it may not be necessary to limit supplementary evaluation of important functions to animal studies, such as repeated-dose toxicity studies lasting up to 14 days. Rather, it would be preferable for the ICH S6 guideline to recognize and encourage the active use of conventional studies in vitro, in particular novel evaluation approaches, including NAMs such as MPS-based systems in vitro and evaluations in silico.

Conclusion

Although some differences remain among geographic regions regarding the definition of NAMs and their regulatory positioning, the overall direction of using NAMs to achieve safety evaluations with greater human relevance, in light of the scientific limitations of animal studies, is shared among the United States, Europe, and Japan [4, 10, 11, 16]. Accordingly, a gradual transition is anticipated from the conventional framework premised on animal studies toward an integrated WoE approach that incorporates multiple sources of information, including NAMs. The ICH S6 guideline already provides a flexible framework based on case-by-case evaluations, and the concepts presented in this article are consistent with this fundamental philosophy. We hope that the perspectives presented here will contribute to future discussions on nonclinical safety evaluations and to the future revision of the ICH S6 guideline.

Funding

This research was supported by grants from the Japan Agency for Medical Research and Development (AMED) under Grant Number 26mk0121281j0003.

Conflict of Interest

The authors declare no conflicts of interest.

Acknowledgments

We thank Mr. Daisuke Sato of the PMDA for his valuable advice on this research project. The views expressed in this article are those of the authors alone and do not represent the official views of the PMDA. We thank Robin James Storer, PhD, from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript.

References

• 1.  International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) ICH S6(R1): Preclinical safety evaluation of biotechnology-derived pharmaceuticals. 2011. Available at: https://www.ich.org/page/safety-guidelines.
• 2.  Science Board to the FDA NAMs Subcommittee. Potential approaches to drive future integration of new alternative methods for regulatory decision-making. U.S. Food and Drug Administration; October 2024. Available at: https://www.fda.gov/media/182478/download.
• 3.  U.S. Food and Drug Administration (FDA). FDA announces plan to phase out animal testing requirement for monoclonal antibodies and other drugs. FDA Newsroom−Press Announcements; April 22, 2025. Available at: https://www.fda.gov/news-events/press-announcements/fda-announces-plan-phase-out-animal-testing-requirement-monoclonal-antibodies-and-other-drugs.
• 4.  U.S. Food and Drug Administration (FDA). Roadmap to reducing animal testing in preclinical safety studies. U.S. Food and Drug Administration; 2024. Available at: https://www.fda.gov/media/186092/download.
• 5.  Pharmaceuticals and Medical Devices Agency (PMDA). Review reports. Available at: https://www.pmda.go.jp/.
• 6.   Chien,  H. T.,  Prior,  H.,  Andrews,  L.,  van Aerts,  L.,  Cauvin,  A.,  Clarke,  D. O.,  Datta,  K.,  Dempster,  M.,  Dybdal,  N.,  Freebern,  W.,  de Haan,  L.,  Herzyk,  D.,  Hey,  A.,  Kissner,  T.,  Kronenberg,  S.,  Leach,  M. W.,  Lee,  D.,  Schutte,  K.,  Sewell,  F.,  Trouba,  K.,  Ulrich,  P.,  Weir,  L. and  van Meer,  P. 2023. Re-evaluating the need for chronic toxicity studies with therapeutic monoclonal antibodies, using a weight of evidence approach. Regul. Toxicol. Pharmacol. 138: 105329.
• 7.   Eastwood,  D.,  Findlay,  L.,  Poole,  S.,  Bird,  C.,  Wadhwa,  M.,  Moore,  M.,  Burns,  C.,  Thorpe,  R. and  Stebbings,  R. 2010. Monoclonal antibody TGN1412 trial failure explained by species differences in CD28 expression on CD4+ effector memory T-cells. Br. J. Pharmacol. 161: 512–526.
• 8.   Craig-Meyer,  D.,  Hollenbaugh,  J. A.,  Morgado,  S.,  McGee,  K.,  Perkins,  E.,  Yarzabek,  B.,  Lapinski,  P.,  Rowse,  A.,  Cooper,  C.,  Fortunato,  M.,  Cocco,  M.,  Cadwallader,  K. and  Munday,  J. 2025. Immunophenotypical characterization of immune checkpoint receptor expression in cynomolgus monkeys and human healthy volunteers in resting and in T-cell stimulatory conditions in vitro. J. Immunotoxicol. 22: 2462106.
• 9.   Rana,  P.,  Hollingshead,  B. and  Mangipudy,  R. 2024. Rethinking the necessity of long-term toxicity studies for biotherapeutics using weight of evidence assessment. Regul. Toxicol. Pharmacol. 153: 105710.
• 10.  Pharmaceuticals and Medical Devices Agency (PMDA). Review report for Reblozyl (luspatercept). Tokyo, Japan; 2024. Available at: https://www.pmda.go.jp/drugs/2024/P20240112001/670605000_30600AMX00021_A100_1.pdf(in Japanese).
• 11.  U.S. Food and Drug Administration (FDA). Monoclonal antibodies: streamlined nonclinical safety studies. draft guidance for industry. December 2025. Available at: https://www.fda.gov/media/189920/download.
• 12.  European Medicines Agency (EMA). Reflection paper on non-human primates in safety testing of human medicinal products and opportunities for 3Rs implementation. EMA/CHMP/55697/2025, 6 October 2025.
• 13.   Rossi,  G.,  Manfrin,  A. and  Lutolf,  M. P. 2018. Progress and potential in organoid research. Nat. Rev. Genet. 19: 671–687.
• 14.   Ingber,  D. E. 2022. Human organs-on-chips for disease modelling, drug development and personalized medicine. Nat. Rev. Genet. 23: 467–491.
• 15.  National Academies of Sciences, Engineering, and Medicine. Nonhuman primate models in biomedical research: state of the science and future needs. Washington, DC: The National Academies Press; 2023. .
• 16.  Pharmaceuticals and Medical Devices Agency (PMDA), Center for Product Evaluation. Thoughts on the use of weight of evidence approach in nonclinical safety evaluation (early consideration), 24 October 2025. Available at: https://www.pmda.go.jp/files/000278904.pdf.