Animal Study Translation: The Other Reproducibility Challenge

Abstract

Animal research is currently an irreplaceable contributor to our efforts to protect and improve public health. Its relevance, importance, and contributions are represented in historical precedent, regulatory expectations, evidence of our rapidly developing understanding of human health and disease, as well as success in the development of novel therapeutics that are improving quality of life and extending human and animal life expectancy. The rapid and evolving success in responding to the current COVID pandemic significantly supported by animal studies is a clear example of the importance of animal research. But there is growing interest in reducing our dependence on animals and challenges to the effective translation of current animal studies to human applications. There are several potential contributors to gaps in the translatability of animal research to humans, including our approaches to choosing or rationalizing the relevance of a particular animal model, our understanding of their biological variability and how that applies to outcomes, the data we collect from animal studies, and even how we manage the animals. These important contributors to the success of animal research are explored in this issue of the ILAR Journal.

animal research, irreproducibility, 3Rs, SARS CoV2, translation

Contemporary societies have generally and rightfully frowned on primary interventional human experimentation complicating our ability to probe the intricacies of human biology and disease. Accordingly, animal studies have for centuries been an important source of study and insight into the human condition. Notably, conduct of animal experimentation prior to human experimentation has been considered an ethical imperative as outlined in the Nuremburg Code of 1946 that was an outcome of the medical atrocities of World War II: “The experiment should be so designed and based on the results of animal experimentation and a knowledge of the natural history of the disease or other problem under study that the anticipated results will justify the performance of the experiment.”^1,2

For centuries, animals have been used as surrogates or models of human biology and pathobiology. Kinter et al chronicle the long and close relationship of humans and animals with significant dependencies by humans on animals originating in a need for food and clothing and evolving to include more scientific dependencies to support efforts to prevent disease and improve health.³ Our current confidence in the relevance of animal studies for insights into human biology comes from observation of morphologic and functional similarities and our fundamental confidence in the evolutionary conservation of basic mammalian and even broader animal biology. Indeed, there is significant homology across the biological spectrum between many higher mammals and humans at the gene, cell, tissue, and organ system level, with recapitulation of most major processes at both the molecular and macromolecular levels. Relatedly, there are even homologies that extend beyond mammals and can be observed at very fundamental levels throughout the animal kingdom.^4–6 Accordingly, animal modeling systems extend far beyond usual laboratory mammal species. The use of nonhuman animal species as surrogates for understanding human biology has also been validated by significant insights into fundamental human physiology developed by the likes of Claude Bernard, the French physiologist, who advocated for animal research as a foundational capability in experimental medicine in his book, Introduction a l’etude de la Medecine Experimentale (1865), and August Krogh, a distinguished Danish physiologist and recipient of the 1920 Nobel Prize for Physiology and Medicine for his work on capillary function, which he studied in frogs.^7,8

A broad variety of animal species has been and continues to be used as “models” of human-relevant biology. These animals are used to interrogate a spectrum of interests, including fundamental biological processes, gene–phenotype relationships, responses to injury, and even to support the development of human medical interventions such as diagnostics, disease prevention strategies, and therapies. As biomedical research has significantly expanded, so has our use of animals to support that research. Use of those animals has been closely monitored and even regulated to ensure their most judicious, ethical, and relevant use.³

A significant source for our confidence in animal systems aside from our understanding of evolutionary conservation is the ability to morphologically or phenotypically replicate biological conditions that we are accustomed to seeing in human clinical medicine; that is, we can phenotypically “translate” them to the human. Our confidence in the animal systems and their relevance to human patients is so entrenched in our approach to studying human biology that they are an expected and foundational component of a translational research effort and have even become a regulatory requirement for some purposes (eg, safety assessment of new drugs and some chemicals). Regulatory agencies such as the US Food and Drug Administration and the Environmental Protection Agency have guidance and requirements that codify the belief that data derived from animal studies are critical to protecting humans from the toxic health effects of environmental chemicals and novel therapeutics.^9,10 Most recently, the role of animal studies in supporting the development of COVID 19 vaccines is recognized in US Food and Drug Administration Guidance.¹¹

Though observational research including the study of organisms across the animal and plant kingdoms and even microbiota continues to be a source of insight into biology generally, biomedical research has become much more oriented toward experimental research using biological systems—including animal systems—that are manipulated to artificially replicate or model human-relevant biology or response to injury. Accordingly, the focus on animal modeling systems has narrowed to a smaller number of animal species, where rodents, particularly mice, have emerged to become most used.³ Though mice are reasonably considered relevant biological surrogates for humans, the motivations for their experimental preeminence has less to do with their biology than their size, cost, and fecundity and our ability to genetically modify them. This approach may be contributing to translational weaknesses in our overall research endeavor.

There is no better or more contemporary an example of how animal research has significantly supported our efforts to manage human disease than that of the rapid response to SARS-CoV-2 and the Coronavirus-induced Disease (COVID) pandemic we have been navigating for over a year. The unexpected emergence and rapid spread of the virus and consequential clinical disease have demonstrated the critical role of animal models in responding to human health threats. The ability to “humanize” mouse models through genetic modification as well as models with natural susceptibility to SARS-CoV-2 such as hamsters, ferrets, and nonhuman primates has provided a portfolio of research tools that have allowed us to gain insights into this unique agent and disease at an unprecedented pace. Significant experience with a variety of animal models used to study other respiratory and systemic viral infections was rapidly leveraged to initiate fundamental disease and therapy development research, which has helped mitigate the profound impacts of this disease. Our characterization of the comparative biology of the human condition and the relevant animal models has enabled our ability to optimize the translational relevance of these animal studies. Zeiss et al and Veenhuis et al provide a thorough review of SARS coronaviruses, the benefits and limitations of individual animal models, contemporary understanding of the human disease, and the comparative immunology underpinning certain aspects of the coronavirus infection and disease.^12,13 Relatedly, Kuiper et al share their effort to model the evolving susceptibility of mice to infection by an evolving SARS-CoV-2 virus to ensure that the animal models used for ongoing research are optimally aligned for the questions being asked of them. Relatedly, they structurally model putative mutations that may enhance human to animal transmission, creating the opportunity for wild or domestic animal reservoirs that could further complicate this persistent pandemic.¹⁴

The biomedical research community and even the general public recognize that animal research is justified by its benefits to humans and even animals through advancements in our understanding and ability to protect and improve human and animal health. Ethical and welfare assessments conducted by legally mandated institutional review committees ensure the merit of that justification for individual studies. These reviews aim to ensure an optimal balance between harm to the animal and the benefit of the research to humans or animals. Integral to ethical review are the fundamental principles of the 3Rs (replace, reduce, refine) articulated by Russell and Burch in 1959, which have also significantly influenced the practice of animal research and supported great progress in the field of laboratory animal medicine.¹⁵ These principles, aimed at ensuring the most judicious use of animals in experiments in which their pain and distress are minimized, have positively progressed our ethical use of animals as surrogates for experimenting in humans. But our commitment to the 3Rs has also influenced our choice of animal species favoring lower sentient organisms or those with less public sensitivity as experimental subjects over potentially more relevant species such as nonhuman primates.¹⁶ It is possible that our commitment to the welfare of animals has negatively impacted the human relevance of our animal research, creating an ethical paradox. The analytical rigor and translational relevance of animal studies are key considerations not always best considered by review processes more focused on ethics and welfare.¹⁷ Current practices would benefit from the addition of a more formalized and structured scientific peer review of animal model selection and study design prior to ethical review. The animal research community would also benefit from sharing animal model experiences and best practices that would distribute knowledge among animal investigators and provide guidance for ethical review committees as they consider the scientific merits of an animal experiment.¹⁸

Ethical review processes are well established and guided by regulatory expectations in most regions of the world. Rigorous and standardized approaches with appropriate oversight and governance ensure that methods to minimize animal use and reduce pain and distress are consistently applied. What is less well-standardized is an approach to ensuring that an animal study will optimally translate to humans when that is the intent of the study. Several factors contribute to that likelihood, including our fundamental understanding of the human disease or biology of interest, alignment of the systems-level physiology, previous experiences with translation, and the reproducibility of responses in the model. Storey et al explore these factors and propose a structured framework that could form the basis of a scientific peer review of the relevance of an animal model prior to ethical review.¹⁹

Despite our long and, arguably, successful history of translational animal research, there is growing interest in decreasing our dependence. These concerns have gained considerable momentum with the emergence of evidence that some animal research is poorly reproducible and often does not replicate in humans. The irreproducibility of animal studies has been highlighted by a few pharmaceutical groups who attempted to reproduce outcomes of preclinical studies, including animal studies reported in the peer-reviewed scientific literature.^20–22 This irreproducibility quickly gained focus as a potential source of drug development attrition, which has also been a growing source of concern because it contributes to the rising cost of new medicines.

Revelation of a reproducibility challenge in applied animal research has fueled 2 opposing efforts: a call for improved animal study reporting standards and advocacy for eliminating animal research. The former has prompted the development of standard reporting practices as represented in the National Centre for the Replacement, Refinement and Reduction of Animals in Research (NC3Rs) ARRIVE Guidelines.²³ The latter has contributed to the animal research ethics debate and put more focus on advancing non-animal modeling capabilities as an alternative. Despite significant progress in non-animal modeling systems of increasing biological complexity, there are no non-animal modeling capabilities that come closer to reproducing the complexity of the human system than animals. This is even though many of the newer modeling systems are incorporating human cells in increasingly complex and in vivo–relevant architectures, increasing their potential for human relevance in very narrow biological scopes.^24,25 Ironically, many of those systems are being qualified by comparing them with outcomes in traditional animal systems. As an example, Huh et al used a mouse perfusion/ventilation model as a comparator to support the in vivo relevance of a unique lung-on-a-chip model in which they modeled interleukin-2 (IL-2)–induced pulmonary edema.²⁶ That approach to qualification of non-animal systems is likely useful but will benefit from animal systems that truly reflect human-relevant biology to ensure that the in vivo relevance is actually to the species of interest. Non-animal modeling systems will rightfully gain significant traction in translational research but are unlikely to replace all animal research in the near term. Paradoxically, translationally relevant animal research may, in fact, be a necessary enabler for a future less dependent on animal research. Currently, our best opportunity for qualifying novel in vitro systems for their ability to model in vivo outcomes is to relate in vitro animal systems to in vivo animal outcomes.

Improving reporting standards for animal studies will improve the reproducibility of those studies and contribute to our judicious and impactful use of animals. Ensuring that those studies model human-relevant biology and conditions optimizing the likelihood of the outcomes replicating in humans is a somewhat different challenge that has received less direct attention and action. As noted above, animal model selection has often been prompted more by economics and technical efficiency than human relevance. There are no established standards for justifying the human relevance of a particular animal model and no established process for challenging that justification as there is for challenging approaches to minimizing pain and discomfort with an ethical review.²⁶

There are several translational issues that could be contributing to our difficulty in consistently replicating animal study outcomes in humans. First and foremost is the inherent variability of biology within and between species. It is an accepted tenet that individual humans do not completely replicate other individual humans, so it makes sense not to expect that individuals of another species will completely replicate other members of that species, let alone to completely replicate human biology. Certainly, there are lots of differences between strains of individual animal species with respect to both normal physiology and response to injury.^27–31 Accordingly, there should be some limit to our expectations for how well an animal or animal study should replicate a human outcome. But, if we cannot mitigate that challenge by optimizing our questions, our choice of animal model, our study design, and our human extrapolation, one would have to question the usefulness of doing the study to begin with.

Though most, if not all, animal species share biological and pathobiological similarities to humans at some level widely exploited in biomedical research, each species has its unique phenotypic features. Recognizing these unique features of normal and abnormal biology is critically important for interpreting outcomes and understanding the implications of these features in the translation of the model to the human condition. Cooper et al, Mangus et al, and Helke et al collectively provide an important and useful overview of background lesions and conditions (ie, abnormal biology) in a wide variety of laboratory species.^32–34 Additionally, Bau-Gaudreault et al provide an overview of key differences and resources for clinical pathology measures for a number of laboratory animal species.³⁵ A thorough understanding of these biological variables inherent in common laboratory animals is essential for considering their impact on the outcomes of a study and the interpretation of experimental observations.

Biomedical research is increasingly reductionist, interrogating the fundamental mechanisms of biology at ever higher levels of molecular resolution. That evolution has not often been accompanied by a similar level of molecular characterization or qualification of the animal models we are using outside of assessing gene sequence homology and expression. It is well recognized that no one model—animal or otherwise—can fully recapitulate the complexity of a human disease. Accordingly, a characterization of potential models at the biological level in which the model is being used is critically important. The breadth of technologies we use to characterize an experimental outcome can also be used to prospectively characterize or phenotype the model prior to the experiment to ensure its relevance to the “human” question to be considered in the experiment.^36–39

Though molecular characterization of common models is improving, that might be at the expense of our appreciation for contextual physiology at the organ system level that no doubt has a profound influence on how a biological system functions or responds to perturbation at the molecular level. For example, cardiac excitation-contraction coupling is a well-conserved cellular function at even the mechanistic level, with many of the same molecular mediators shared across species with rhythmically contracting hearts.⁴⁰ No doubt there are many animal species that could be used to tease out the functional mediators of cardiac contraction. But characterizing the response to perturbation or pharmacologic modulation of that system is likely very different in a rat with a resting heart rate of 350 beats per minute versus a resting human rate of approximately 70. The quantitative dynamics of those 2 systems are different and that difference is influential. That influence probably contributes to the challenges of developing drugs that effectively mitigate the progression of human heart failure.⁴¹

As physiologic measures are commonplace in both human and veterinary clinical settings, they should be commonplace in experimental animal research. No doubt technology has been limiting in our ability to collect meaningful physiologic data in “free-ranging” animals and humans without restraint or surgical manipulation, but those limitations are quickly waning.^42–44 Wearable biomonitoring devices in people are allowing us to be more aware of and involved in our personal health.⁴⁵ Continuous measures of health parameters supported by these devices will enable us to identify disease earlier in its progression and also redefine how we consider the continuum of health and disease. Likewise, emerging digital vivarium technologies are supporting more continuous assessment of animal health but also improving our environmental data collection and integration into the experimental outcomes.^46,47 Defensor et al explore how advances in digital technology are not only enabling the collection of experimental metadata improving reporting of animal studies but also integration of biosensors into home-cage environments that measure fundamental physiologic endpoints as well.⁴⁸ Parallel developments in biosensor technology for both humans and animals will likely lead to the next revolution in the way we consider and characterize health and disease, allowing us to consider them in the dynamic contexts in which they present and progress, identify chronic progressive diseases much earlier in their progression, identify patients most at risk of clinical disease, and improve the translational relevance of animal studies.

Related to that, the primary intent of a study will likely influence the rigor with which translational relevance is considered for a particular animal model. Kimmelman et al usefully categorized experimental studies as either “exploratory” or “confirmatory.”⁴⁹ Exploratory studies might be considered as hypothesis generating and primarily focused on “discovering” a novel insight or mediator of a biological process. In the context of translational research, the hypothesis generated is likely to be considered a “human” one requiring “confirmation” in a relevant surrogate for the human or by a human observation. Again, referencing back to the cardiac excitation–contraction coupling biology, a rodent could be a useful surrogate for the human heart for understanding key events and mediators in the excitation–contraction cycle. Quantitatively extrapolating or confirming the role of those events and mediators or, more particularly, the effects of modulating those events and mediators should use an animal species with a more human-relevant cardiac physiology and therefore a better “fit” for the question.⁴⁶ Accordingly, rodents are often used in early drug safety–screening studies, but the dog is considered the definitive species in drug development to characterize any potential functional cardiovascular liabilities prior to human trials.⁵⁰ The translational relevance of cardiovascular studies are further enhanced through the use of “translational biomarkers” or measures that can be collected in both dogs and humans, such as heart rate, rhythm, contractility, and blood pressure.

Some of the biological differences in traditional animal models are iatrogenic and induced by our husbandry and study design practices. It is well recognized that the room temperature maintained in most animal facilities that is most comfortable for human caretakers is well below the euthermic temperature for rodents. Consequently, laboratory rodents exhibit a significant upregulation of sympathetic autonomic tone, increasing most parameters of cardiovascular function and metabolism while decreasing immune responses.^51–53 Use of nesting materials and group housing is intended to mitigate some of those effects but do so incompletely.⁵² Likewise, efforts to maintain “clean” facilities to control animal colony disease have an unintended consequence in that laboratory rodents have immature immune systems, complicating the translation of their immune responses to humans.^54,55 One of those husbandry practices, particularly for nonrodent species, is vaccination to control communicable diseases. Layton et al share an experience where vaccination to prevent distemper in ferrets caused an unexpected hypersensitivity reaction to SARS-CoV-2 infection.⁵⁶ Likewise, anesthesia is often used in animal studies to enable collection of important biological data in restrained animals while minimizing both their physiological pain and distress. Navarro et al review the art and science of mouse anesthesia to ensure optimal outcomes for both the experimentalist and the animal.⁵⁷

Animal research is a critically important contributor to the biomedical research community’s efforts to protect and improve human health. The privilege we have to conduct that research is being challenged and rightfully so. For the benefit of humans and to justify our harm to animals, it is important that we optimize our use of animal research for human relevance and application. That includes using well-characterized animal models that are appropriately aligned to the human biology of interest, that we understand their normal biological variability and how it relates to the endpoints of interest, that we think holistically about the animal subjects in our studies, that we measure relevant endpoints, and that we rigorously challenge our justifications for their use. Maybe our ability to model human-relevant biology and disease without animals will continue to improve and we will see a day in biomedical research where animal research is not necessary. In the interim, it is imperative that we ensure judicious use of appropriate animal models in appropriately and rigorously designed animal studies.

This thematic issue of the ILAR Journal will explore the full spectrum of comparative and translational biology and pathobiology with an emphasis on how we might challenge traditional approaches to animal studies to ensure they support successful extrapolation or “translation” to humans.

Potential conflicts of interest. No reported conflicts.