Nervous system diseases and disorders are highly prevalent and substantially contribute to the overall disease burden. Despite significant information provided by the use of animal models in the understanding of the biology of nervous system disorders and the development of therapeutics; limitations have also been identified. Treatment options that are high in efficacy and low in side effects are still lacking for many diseases and, in some cases are nonexistent. A particular problem in drug development is the high rate of attrition in Phase II and III clinical trials. Why do many therapeutics show promise in preclinical animal models but then fail to elicit predicted effects when tested in humans?
On March 28 and 29, 2012, the Institute of Medicine Forum on Neuroscience and Nervous System Disorders convened the workshop “Improving Translation of Animal Models for Nervous System Disorders” to discuss potential opportunities for maximizing the translation of new therapies from animal models to clinical practice. The primary focus of the workshop was to examine mechanisms for increasing the efficiency of translational neuroscience research through discussions about how and when to use animal models most effectively and then best approaches for
1The planning committee’s role was limited to planning the workshop, and the workshop summary has been prepared by the workshop rapporteurs as a factual summary of what occurred at the workshop. Statements, recommendations, and opinions expressed are those of individual presenters and participants, and are not necessarily endorsed or verified by the Institute of Medicine, and they should not be construed as reflecting any group consensus.
the interpretation of the data collected. Specifically, the workshop objectives were to
• Discuss key issues that contribute to poor translation of animal models in nervous system disorders.
o Examine case studies that highlight successes and failures in the development and application of animal models.
• Consider strategies to increase the scientific rigor of preclinical efficacy testing.
o Explore the benefits and challenges to developing standardized animal and behavioral models.
o Identify methods to facilitate development of corresponding animal and clinical endpoints.
• Identify methods that would maximize bidirectional translation between basic and clinical research.
• Determine the next steps that will be critical for improvement of the development and testing of animal models of disorders of the nervous system.
ORGANIZATION OF THE WORKSHOP AND SUMMARY
The first session of the workshop reviewed the current state of animal models for the study of nervous system disorders (Chapter 2). This was followed by concurrent breakout group discussions of the application of animal models to a number of specific diseases and conditions, including neurodegeneration, Alzheimer’s disease, stroke, schizophrenia, addiction, and pain. Upon reconvening the full workshop, the moderators for each breakout group summarized discussion highlights (Chapter 3). The next three workshop sessions expanded on the issues raised in the breakout group discussions regarding the utility of standardization of animal models (Chapter 4), corresponding animal and clinical endpoints (Chapter 5), and the translational gap between preclinical and clinical studies (Chapter 6). In the closing session, the main themes from the presentations and discussions were summarized by the session chairs (Chapter 7).
Topics Highlighted During Presentations and Discussions2
Several main issues were discussed across multiple workshop presentations and discussions:
• Utility of animal models: Throughout the workshop, many presentations and participant remarks highlighted the utility of animal models in understanding of fundamental pathology and disease pathogenesis. Animal models have also been an important component of the development pathway for nervous system disorder therapeutics. In particular, animal models have aided in the validation of potential therapeutic targets and assessment of pharmacodynamics. For example, these models have demonstrated that functional components of hippocampal circuits are similar between aged rats and humans. Measurements of prepulse inhibition in animals have been used as a predictive model of antipsychotics in humans due to similarities across species in response characteristics.
• “Animal model” or “model animal”? During the workshop many participants reiterated that there are really no animal models that fully mimic or recapitulate human diseases or disorders. Rather, models of a particular aspect or specific target of interest for any given nervous system disorder have been generated through genetic, surgical, or pharmacological manipulation or through other means in whole animals. While these models cannot replicate every aspect of a complex human disease, they are useful for dissecting out particular mechanisms, confirming hypotheses and developing therapeutics. In this respect, nomenclature was raised as an issue. Several participants noted that improved stratification of models, to match specific components of disease mechanisms or phenotypes, might lead to a greater ability to answer well-defined questions and the eventual development of better therapeutics. Following on this point, there was some discussion about not labeling models as a model of a whole disease or disorder (e.g., an animal model of schizophrenia), but rather to label them for the hypotheses or mechanism they are testing. Some participants noted that the challenge of this approach is that symptoms are usually addressed in the context of
2Rapporteurs’ summary based on the presentations and the summaries presented by the meeting and session chairs in the final workshop session.
the disease and that research and drug development requires a specific labeled indication.
• Standardization versus best practices: Standardization in the context of the workshop was discussed both in terms of development of similar animal testing paradigms across laboratories or the development of a single “standardized” model for a particular aspect of a disease or disorder. In the discussions about standardization, a range of opinions were expressed. Many in the breakout group on addiction, for example, supported the use of multiple models that employ different approaches, so that converging evidence could be gathered, rather than promoting one or two select models. In contrast, some participants in the neurodegeneration breakout group emphasized the need for standardization of animal models of neurodegenerative diseases. Across the different research areas a variety of concerns were raised, one concern was that premature standardization might not be helpful and could stifle innovation by constraining research to specific areas or models. Some participants believe that over-standardization might artificially inflate sensitivity and reduce generalizability, resulting in clinically irrelevant findings.
• Reproducibility of research: Challenges associated with the reproducibility of animal model research were discussed. The standardization session began with a discussion about the increasing difficulty of reproducing published studies and highlighted a recent article that explored this issue (Prinz et al., 2011). Many participants agreed that improving experimental protocols and statistically analyzing experiments appropriately might help in moving the field forward and increasing reproducibility. A few participants noted that increasing the level of information detailed in publications, including the explicit reporting of replication of experimental results, would also be beneficial. Several participants emphasized the importance of reproducibility and the scientific validity of data as decisions are made about when to move from the laboratory into clinical trials.
• External and internal validity: The focus of the workshop was on the utility of animal models in the translation of basic research to human diseases and disorders. Workshop participants spent the majority of the workshop discussing the external validity of animal models or the extent of which research that uses animal models can be correctly generalized to human diseases
(van der Worp et al., 2010). However, as evidenced by the numerous comments by speakers and participants, external validity can only be discussed in combination with internal validity or the extent to which the design and conduct of a research experiment eliminates the possibility of bias (van der Worp et al., 2010). Many participants discussed the importance of randomization of animals to treatment groups, blinding of treatment assignments, sample size calculation, and outcome measures; all these are factors that might negatively impact the internal validity of the experimental results.
• Bidirectional translation: Many participants believed that animal models and human clinical research inform each other through bidirectional translation, from the bench to bedside and back again. Several participants discussed the importance of cross-validation of animal model endpoints with clinical measures and of clearly establishing what the corresponding endpoint is intended to predict. It was suggested that failures in clinical trials might be due to a mismatch between endpoints used in preclinical animal studies and those used in trials. A few participants expressed the view that there might be particular utility in the use of imaging and biomarkers in both animal models and human clinical research. However, it was noted that in some diseases, biomarkers assayed in humans do not have parallels in animals. Some participants suggested that data collected from negative outcomes, either in animal or preclinical studies, might also help to improve translation.
• Collaboration: Collaboration among sectors was discussed as a critical component in efforts to bridge the gap between preclinical and clinical research and to facilitate bidirectional translation. Several examples of precompetitive alliances3 and cross-sector collaborations were discussed, both lateral efforts (across institutions and companies) and vertical (from preclinical through to clinical). Examples included a government-funded consensus building initiative; a government-facilitated, precompetitive public–private partnership designed to address specific issues in drug development; a for-profit consultancy bringing validated models from academia to drug developers; and the use of quanti-
3Precompetitive alliances are collaborations among competitors to achieve goals that can be more effectively accomplished by a group effort and have the potential to benefit all groups involved (IOM, 2010).
tative systems pharmacology as a translation tool, applying mathematical model–based decisions to support drug development. Many participants said that the inclusion of regulators in early conversations around validation and qualification of endpoints might also help to facilitate translation and the eventual development of new therapeutics.
• Communication: Many participants emphasized the need for similar terminology across basic and clinical research to facilitate translation of success in animal models to success in patients. It was noted that cross-disciplinary collaborations already require an understanding of each other’s vocabulary. There was also discussion on the publication of animal studies, particularly concerns about the impact of bias were discussed (e.g., failure to publish negative results, publication of poorly designed or executed studies). Several participants mentioned the need for honesty in discussions about the predictive value of any given model. Several participants noted that in some cases it is not the models that need to be improved, but the dialogue about the models needs to be forthcoming. Of particular note, was the need to establish realistic expectations about what animal models predict.
• Training: Throughout the workshop, issues related to experimental design and statistical analysis of results were discussed as they related to the ability to translate data produced from animal models into potential therapeutics. Many participants agreed that these issues might be resolved with greater emphasis of these topics during training of graduate students and postdoctoral fellows. Several participants expressed concern over the deficiencies that current trainees have in basic research design and statistical analysis which are prerequisites for conducting any experiment. Many participants highlighted specific examples of areas where knowledge of and focused training on might be beneficial, including identification of primary experimental hypotheses and outcome measures; the importance of predefining analyses plans and inclusion and exclusion criteria; completion of sample size calculations; and identification of dependent and independent variables.
• Animal model alternatives: The workshop closed with a spirited discussion about the overall value of animal models as part of the therapeutic development pathway. The decreasing confidence in the ability of animal models to predict efficacy of drugs with
novel mechanisms of action was emphasized by several workshop participants. Concerns were raised about the potential that animal models may, in fact, be screening out potentially effective compounds. In particular, one participant noted that if the mouse models for polygenic psychiatric disorders are really as poor as described, perhaps it is time to ask under what circumstances it would be both worthwhile and ethical to go straight into human clinical trials after establishing safety. Should research continue with the current emphasis on animal models, or should more focus be placed on cellular models and/or Phase 04 human clinical trials? No answer was provided to this question, but many participants agreed that exploring methods to speed therapeutic development to first-in-human trials would require changes in the way current animal models are used. A few participants cautioned that, at minimum, this approach would not be successful without validated targets with the potential for efficacy and clinical benefit. Several participants also suggested that using animal models in conjunction with new emerging tools and technologies might help in the creation of a complete picture of disease pathophysiology and mechanisms, which would better aid in the creation of new pathways for the development of therapeutics.
4An exploratory investigation new drug study, or Phase 0 study, is conducted very early in Phase I and involves microdosing in a very limited number of human participants. See Guidance for Industry, Investigators, and Reviewers: Exploratory IND Studies, http://www.fda.gov/downloads/drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM078933.pdf.