7
Regulatory Perspectives
Regulatory agencies grapple with many of the questions discussed at the workshop regarding what levels of evidence are needed to conduct efficient clinical trials in humans, what constitutes an acceptable outcome measure, and what is the role of surrogate outcomes, said Linda Brady, director of the division of neuroscience and basic behavioral science at the National Institute of Mental Health. The workshop specifically focused on how to address these issues when predictive animal models of disease are not available for preclinical studies.
PERSPECTIVES FROM THE FDA AND EUROPEAN MEDICINES AGENCY (EMA)
Robert Temple, deputy director for clinical science at the Center for Drug Evaluation and Research at the Food and Drug Administration (FDA), noted that Title 21 of the Code of Federal Regulations does not require evidence of effect in an animal model to examine a drug’s pharmacologic effects, efficacy, and toxicity before conducting clinical trials or as part of a marketing application. Rather, it leaves it to the investigator to decide what tools are needed. An Investigational New Drug application, he said, should include the rationale for the drug or research study, which could include animal data as well as pharmacologic and mechanistic information. However, the regulations do not state how persuasive or credible the information must be. What the FDA wants to see from Phase I and II studies, he said, is enough information about pharmacokinetics and pharmacologic actions to permit the design of well-controlled, scientifically valid Phase II and III studies with an appropriate dose. Yet it would be very unusual to “hold” a study because of a lack of preclinical or mechanistic information, with the possible exception of a trial in which patients are denied a known, effective disease-modifying treatment.
Surrogate endpoints, sometimes supported by animal models, have been proposed as a basis for approval of some drugs and have been used for accelerating approval (with a requirement for post-approval studies to support clinical benefit). For example, in considering approval of interferon-based drugs for multiple sclerosis, strong magnetic resonance imaging data supported a single study showing reduced exacerbation rates as the basis for accelerated approval, despite a lack of clinical data demonstrating efficacy. Accelerated approval is generally used for serious diseases for which there are no treatments when a surrogate marker is “reasonably likely” to predict a clinical benefit. A drug does not have to reverse the course of the underlying disease to be approved, said Temple. Many drugs are approved for treating symptoms if the benefit is well documented and outweighs the risks. For example, in multiple sclerosis, drugs have been approved based solely on reduced exacerbation rates. Reducing progressive disability is the desired long-term outcome, and many treatments have been shown to do that, but it is more difficult to document in the time frame of a clinical study. Temple cautioned, however, experience with surrogates shows that sometimes an early effect does not turn out to represent a clinical benefit.
Problems occur when a predictive animal model is unavailable and there is no plausible pharmacologic effect or good biomarker, said Temple. In these cases, it is still possible to design more efficient and rigorous trials, using enrichment measures, and test multiple doses to show a clear dose−response curve. In some cases, only a subset of study participants responds. In such cases, trials can be designed using placebo lead-ins and prognostic enrichment strategies. If an unanticipated responder subset is identified in a randomized trial, apparent responders can often be restudied in a new trial with a randomized withdrawal design. Temple provided a few examples in which small studies of this type were used as part of the basis for approval of neurologic drugs: Xyrem for cataplexy, and tetrabenazine for Huntington’s disease.
Maria Isaac, senior scientific officer for the EMA, said that in Europe as in the United States, when predictive models are not available, it is especially important to understand the pharmacology, the target, and the kind of risks that might be associated with a study. She said a practice has evolved whereby sponsors submit first-in-human and early-phase clinical trials with multiple components for regulatory review. For example, separate components may examine single and multiple ascending doses, different age groups, and early proof-of-concept and proof-of-principle aspects. In addition, European regulators support incorporating biomarkers for enrichment and in silico models for dosing, said Isaac.
NEW APPROACHES IN ESTABLISHING SAFETY AND CONDUCTING TOXICOLOGY STUDIES
Thomas Hartung, professor and chair of evidence-based toxicology at the Johns Hopkins Bloomberg School of Public Health, offered a perspective based on his experience as head of the European Commission’s Center for the Validation of Alternative Medicine, his experience at Hopkins, and his work at Organome, a technology company that is developing “mini-brains” similar to the organoids described by Lee Rubin in Chapter 4.
Hartung said that about 20 percent of drug failures in clinical trials are attributable to unanticipated side effects, and that side effects account for 8 percent of drugs withdrawn from the market. Furthermore, while animal tests for toxicology are typically done using good laboratory practices and internationally standardized (and often validated) protocols, reproducibility remains low. Moreover, there are some instances where
animal toxicology tests have led to incorrect conclusions. Human cell-based systems could potentially mitigate this species-difference problem, he said; however, with regard to reproducibility, cell-culture tests may not be much better than animal studies because many cell cultures are misidentified (Kleensang et al., 2016) and mycoplasma infected (Young et al., 2010). To address these concerns and develop guidance about cell culture practices, the European Science Open Forum created an international good cell culture collaboration (Pamies et al., 2016). Hartung and others have also focused attention on improving the quality of reporting of toxicology studies as a step toward evidence-based toxicology (Samuel et al., 2016; Stephens et al., 2016).
Other approaches, including increased curation of legacy data and the use of high-throughput screening, multi-omics technologies, and high-content imaging, also offer the potential of transforming toxicology testing, said Hartung. For example, the Human Toxome Project,1 funded by the National Institutes of Health Transformative Research Grant program, brought together multiple institutions to formulate multi-omics approaches for developing technologies to better understand molecular pathways of toxicity. Another project in Europe brought together legacy data to build the largest toxicology database in the world (Luechtefeld et al., 2016). In parallel, an international cross-industry, multistakeholder group has been formed to develop new “Good Read-Across Practice” guidance for filling data gaps on the effects of chemicals on humans and the environment (Ball et al., 2016).
Hartung also believes that organotypic cultures such as mini-brains can play an important role in toxicology testing, especially if they can be standardized. The Organome mini-brains, for example, comprise more than five types of neurons, including oligodendrocytes and astrocytes; form circuits; and are electrophysiologically active. Forty percent of the axons are myelinated, indicating a degree of maturity that is difficult to achieve in culture, said Hartung.
Although these mini-brains have not yet identified a single unknown toxicant, they have been used to demonstrate MPTP (N-methyl-4phenyl-1, 2, 3, 6-tetrahydropyridine) toxicity similar to that shown using the MPTP rat model of Parkinson’s disease (PD). They may also be useful to study the effect of viral infection—such as Zika virus—on the brain, said Hartung. Because the mini-brains are derived from induced pluripotent stem (iPS) cells, he said there is also the opportunity to de-
___________________
1 For more information, go to http://humantoxome.com (accessed November 18, 2016).
velop them with the genetic background of an individual with the disease to demonstrate how a genetic risk factor might interact with environmental risk factors, such as a pesticide, to produce disease. As more is learned from various human-relevant models, Hartung predicted that the value of animal models may be reduced; however, he agreed that this transition will take time. Moreover, he stated that investigators and regulators should be open to new approaches and all types of information that may become available in the future.
FINAL REMARKS
Although animal models remain a critical component of basic research, Steven Hyman noted that they are less than ideal for validating targets and predicting efficacy. He suggested that cellular models and three-dimensional neuronal organoids may be the best tools for studying polygenic diseases such as schizophrenia. Nita Farahany concurred, commenting that all emerging technologies in animals, cell systems, organoids, and other types of cultures offer the potential to build the levels of evidence that will enable translation from genes to cells and from molecules to animals, and eventually to well-designed human clinical trials. She said that taking advantage of this potential will require reinvigorated investment in central nervous system (CNS) research and development from biotechnology and pharmaceutical companies. At the same time, increased investment for natural history studies with deep phenotyping will be needed to enable segmentation of heterogeneous populations and testing of specific biological questions. Hyman added that an increasing understanding of genetics and genetic risk factors might also allow stratification in natural history studies, further increasing the power of those studies.
Kim Andersen commented that a translational focus is needed even at the preclinical stage of development to build and strengthen links between what is seen in patients and research efforts in academia. Neither academia nor industry can do this alone, he said, suggesting that new types of research structures and public−private partnerships are needed to ask key questions in a cross-disciplinary way. Hank Greely, director of the Stanford Program in Neuroscience and Society at Stanford University, added that one of the problems is that there is no entity or organization dedicated to the promotion of human trials when predictive models
are unavailable. One thing such an entity could do is publicize studies that achieved success in the absence of predictive models.
Rita Balice-Gordon suggested that charting a path forward will require additional discussion about companies’ risk tolerance and perhaps expanding the definition of risk, recognizing that there will be an expanded set of unknowns associated with future human trials. Moving forward into an era of greater uncertainty will also, by necessity, require more data sharing, she added. Several workshop participants said the absence of a clear regulatory path is a barrier to developing treatments for chronic diseases, where combinations of surrogate markers may be used as the basis of approval, even if it is provisional approval with postmarketing monitoring in Phase IV.
A few participants also expressed concern about having an adequately trained workforce, particularly in basic science, to ensure continued progress in developing therapeutics for nervous system disorders. People who sit at the interface between preclinical and clinical science are vanishing from the workforce, said Frances Jensen, and the pressure on basic scientists to move toward translational and clinical projects that may be poorly equipped to conduct has exacerbated this problem.
In his closing remarks, Hyman said that while there is a “certain amount of sober realism” about how well the CNS models in development translate to the human condition, he remained optimistic because, rather than “hanging on to the old way of doing things, people are thinking creatively about how to go forward.”