Molecular Medicine Approach

Bear began by outlining the promise of molecular medicine, an approach that starts with a disease, proceeds to find its cause, and teases apart how the cause exerts pathophysiological effects by conducting basic research. This process results in a mechanistically inspired identification of a molecular target. After validation of the target succeeds, novel drugs aimed at the target can be tested first in animal models and then in clinical trials.

The molecular medicine approach has successfully led to ongoing clinical trials for Fragile X syndrome, which is the most widespread single-gene cause of autism. The genetic cause, reported in 1991, was found to be a transcriptional silencing of the Fragile X Mental Retardation 1 (FMR1) gene, resulting in absent expression of the FMR1 protein (FMRP). By 1994, the first FMR1 knockout mouse was made. Eight years later, in 2002, research on basic neurobiology revealed that knockout animals showed an abnormal type of synaptic plasticity marked by exaggerated function of the metabotropic glutamate receptor 5 (mGluR5). This key finding was validated in 2007 by the demonstration that numerous Fragile X phenotypes are corrected in FMR1 knockout mice by genetic reduction of mGluR5 protein production (Dölen et al., 2007). With the target validated, animal studies and clinical trials ensued with inhibitors of mGluR5.

This line of investigation led to the mGluR theory of Fragile X, which holds that many of the syndrome’s psychiatric and neurological abnormalities are a downstream consequence of exaggerated mGluR5 activation (Bear et al., 2004). The normal FMRP is an mRNA-binding protein that functions at many synapses to repress local protein synthesis stimulated by mGluR5 (Bhakar et al., 2012).

One part of the Fragile X story worth noting is that it was a “community” effort: 40 distinct phenotypes were catalogued in knockout animals, including flies, zebrafish, and mice, in at least 31 different publications (Bhakar et al., 2012). The lessons learned from the Fragile X story, according to Bear, are (1) defined genetic etiology led to valid animal models; (2) fundamental synaptic biology revealed signaling defects; (3) core pathophysiology is evolutionarily conserved; (4) disease-modifying treatments are feasible; (5) treatments can be successful after symptom onset; and (6) multiple mutations may converge on a common signaling pathway.

Bear was careful to point out the risks incurred at every stage of the process, most notably in the design of clinical trials of mGluR5 inhibitors. He observed that fulfilling the promise of molecular medicine requires many educated guesses about the following:

• Drug formulation
• Patient selection (e.g., age, ratings at baseline, etc.)
• Dose selection
• Treatment duration
• Number of subjects
• Number of clinical sites
• Choice of endpoints
• Statistical analysis plan
• Agreement with the U.S. Food and Drug Administration (FDA) on all of the above
• Training of sites in Good Clinical Practice so that data are acceptable to FDA

Experimental Medicine Approach

The most challenging step in the drug development process, Krystal said, is the translation of cellular or animal models to healthy human subjects and then to patients. Krystal noted that researchers cannot go directly into the patient population without knowing whether the biology of the animal models applies to healthy humans. The drug development process is exploratory; it is an iterative process of testing specific mechanistic hypotheses across cellular and animal studies, healthy human subjects, and clinical trials, Krystal said.

The experimental medicine approach, according to Krystal, involves the ability to evaluate the effects of a putative therapeutic agent in humans and examine outcomes at various levels (e.g., behavior, cortical activity, and brain imaging). The goal is to produce therapeutic change in a pathophysiologic process, quickly, and in a laboratory setting. This approach uses drugs as clinical probes in order to identify or verify a target and/or study mechanisms of disease. Similar to proof-of-concept clinical trials, this approach examines the ability of a drug to act on a pathological target and affect a biological endpoint as measured by a biomarker. In general, the experimental medicine approach relies on human studies rather than animal models (Insel, 2012b).

The approach could be useful in several ways to advance and accelerate drug development, said Krystal. First, for go/no-go decisions, using positron emission tomography (PET) imaging, researchers can determine whether a drug enters the brain and reaches the target. Second, the experimental medicine approach can increase or decrease confidence in a particular drug, especially in Phase II. Proof-of-mechanism and proof-of-principle studies could be conducted using this approach to accelerate the development process for several reasons:

• Studying healthy subjects in pathology models instead of patients is cheaper and quicker.
• Biology readouts can be based on short-term rather than long-term exposure (limited toxicology).
• Human pharmacology studies might often be more tractable than studying the disorder because researchers can reduce heterogeneity and the impact of placebo response while probing specific mechanisms.
• Experimental medicine approaches model pathology, which is often necessary to see a therapeutic drug effect.

Krystal used this approach when deploying the drug ketamine to probe pathological mechanisms underlying schizophrenia. Krystal and colleagues demonstrated that ketamine, if given to healthy subjects, produces both positive symptoms of schizophrenia—hallucinations and delusions—and negative symptoms—withdrawal, apathy, and lack of motivation (2005). The schizophrenia-like effects of ketamine are not altered by medications such as benzodiazepine, lorazepam, or two antipsychotic drugs, haloperidol and olanzapine. Findings suggest that there are a variety of biological mechanisms underlying schizophrenia, and the ketamine model, in particular, is related to glutamate circuit dysfunction.

Pharmacologic studies drive translation from animal to human to patient, noted Krystal, and can be useful for asking specific questions that inform the drug development process. Still, it is important to point out that the experimental medicine approach, particularly with regard to ketamine, is designed to probe specific mechanisms, not the full biology, of the disease. Krystal concluded that pharmacologic studies might be used in a thoughtful and hypothesis-based way to inform the testing of specific therapeutic mechanisms in an iterative process that could to lead to drug discovery efforts.

The molecular and experimental medicine approaches are just two ways to improve drug development by rethinking conventional methodologies. Examining the pathophysiology of diseases and changes that occur from drug interactions in humans may improve translation from preclinical to clinical studies. In addition to the use of novel approaches, Krystal noted paradigm changes to the current drug development pipeline are needed to facilitate a more efficient process.

New Scientific Utility Criteria

Karlawish said that the scientific community might opt for revised scientific utility criteria, namely, diversity, competition, degrees of validity, innovation, and value to patients in the selection of future tools and technologies. Karlawish provided a historical overview of the use of the inbred mouse and discussed how new tools and technologies might be developed to reflect the current needs of the field. Karlawish opened by describing a current trend in translational failures: the inability of some drugs that show promise in preclinical animal models to translate into successful therapeutics. Another way to frame this problem is that the inbred mouse, which dominates drug discovery, has not had a commensurate return on investment. This problem is not unique to the neurosciences, said Karlawish, and holds true in most other fields of medicine, especially cancer and infectious disease.

For more than 70 years, the inbred mouse has dominated science. Karlawish noted that the inbred mouse, in particular, was chosen as “a matter of policy” due to its small size, easy maintenance, and low cost. During its initial use, the underlying scientific utility or “value” of the inbred mouse included efficiency, standardization, collaboration, and communication among laboratories, and its ability to produce highvolume data and increase fundamental knowledge. Today, roughly 60 percent of animals used for research protocols are mice (European Commission, 2010). Although inbred mice have substantially increased knowledge, the benefits of their use need to be weighed against potential risks—namely, premature standardization, which can stall progress, said Karlawish.

Heavy reliance on the inbred mouse might be precluding the use of a better tool or technology for generating new knowledge and treatments, noted Karlawish. The inbred mouse was not necessarily the wrong choice for the past 70 years, but rather, a choice that reflected scientific and society needs at the time, he continued. Although, there are several benefits to animal models, the inability for some drugs that show promise in preclinical studies to translate to clinical trials is still an issue. Commonly used animal models, such as the inbred mouse, do not always fully recapitulate human diseases, and therefore may limit the array of questions that can be answered (Bolker, 2012). Karlawish reiterated this point, stating that when the field is not sure where to go, it may be beneficial for researchers to be open to collegial competition and diversity. Researchers might consider utilizing other models that are also aligned with their specific research questions, taking into account environmental and other external factors (Bolker, 2012).

New Paradigm for Drug Screening

Brandon spoke about the need for a paradigm shift in screening nervous system drugs. As an example, Brandon used the conventional drug-screening paradigm for producing non-innovative, “me-too” compounds, which are pharmacologically similar to older drugs, but have reduced side effects (Rizzo et al., 2013). Conventional screens begin with high-throughput screening of large chemical libraries, resulting in identification of lead compounds for which in vivo screening and assessment of efficacy is conducted. Those steps are followed by demonstration of efficacy relative to standard of care, determining side effects, and extended characterization. But this paradigm has been largely unsuccessful at finding novel targets and mechanisms because the targets studied are generally not relevant to disease, Brandon noted.

Brandon outlined components of a potential new drug-screening paradigm that might produce greater success (see Table 5-1). Additional components include

TABLE 5-1

Paradigm Shifts Aimed at Improving Identification of Potential Therapeutics.

• target modulation to demonstrate a pharmacological effect, which is typically a behavioral effect, not necessarily related to efficacy, at doses that occupy the target;
• evidence for efficacy potential using a relevant perturbation in a disease model;
• modulation of physiology endpoints;
• side-effect profiling; and
• screening to identify drugs whose effects may translate across assays.

Finally, Brandon suggested that understanding why drugs fail and then repeating the experiment may help in avoiding the abandonment of new drugs (Rizzo et al., 2013). Brandon’s experience with an inhibitor of phosphodiesterase 10A (PDE10A) exemplifies the limits of the conventional paradigm for drug screening. PDE10A is expressed primarily in medium spiny neurons of the striatum. Because PDE10A degrades cyclic adenosine monophosphate (AMP), its inhibition results in a robust elevation of AMP. The PDE10A inhibitor showed acute efficacy in animal models of schizophrenia and Huntington’s disease (Brandon et al., 2008; Giampà et al., 2010). Efficacy was shown in a stimulated locomotor assay and in other assays used in conventional drug screening paradigms. Despite showing efficacy in multiple conventional screening assays, the drug failed to show efficacy in Phase II trials in schizophrenia. Had investigators evaluated the drug’s potential according to Brandon’s suggested new paradigm, it might not have passed the screen. The drug showed target engagement and target modulation, but no evidence of reversal of positive symptoms, negative symptoms, or cognitive deficits in schizophrenia. It also did not show modulation of physiology endpoints in positive symptoms, cognitive symptoms, and brain network activity. Failure in the new screening paradigm, in other words, may explain why the drug did not work in human clinical trials. Brandon and colleagues, alongside the National Institute of Mental Health (NIMH), are considering more experiments to determine specific reasons for the failure and whether target occupancy in healthy volunteers and patients might be a critical component.

The lesson Brandon learned from the failure of the PDE10A inhibitor clinical trial is that investigators cannot rely on the conventional paradigm for drug screening. The PDE10A inhibitor showed promise in preclinical studies and Phase I trials, but failed in Phase II. Due to this experience, Brandon suggested a new drug screening paradigm with more rigorous components that could be beneficial to accelerate the drug development process (see Table 5-1). Brandon concluded that the field has misinterpreted animal data, oversold the same data, and has been reluctant to do experiments in humans. To develop therapeutics, researchers need to be willing to do the difficult experiments in both preclinical models and in an experimental medicine setting and accept some loss of speed (see Leaf, 2013).

Reversing the Pipeline

As previously highlighted in Chapter 2 by William Potter, senior advisor in the Office of the Director of NIMH, the current development pipeline moves from cellular/molecular mechanisms to animal models to control populations, and finally to the patient population. Several speakers noted that reversing the experimental pipeline by starting in humans for target identification and validation, then moving to animal models for target engagement, may improve the drug development process.

Combination Therapy

Austin highlighted the value of combinatorial strategies in drug development. Combinatorial strategies have become popular in diseases where the target cell mutates, such as in cancer and infectious disease. But the nervous system is adaptable in a more rapid way: it can adapt on a scale of microseconds. Therefore, it should not be a surprise that the nervous system is resistant to interventions, said Austin. When monotherapy does not work, perhaps the conclusion should be that the system adapted, rather than the more typical conclusion that the target was not valid. Combinatorial strategies should be considered in nervous system disorders for these reasons, reiterated Austin.

Changing scientific approaches to the drug development pipeline to address current challenges in the field may be beneficial to the development of therapeutics for nervous system disorders. Several speakers stated that changes to the current paradigm are needed to facilitate a more efficacious drug development pathway to include developing new tools and technologies that are based on scientific utility criteria desired from the field today; shifting the drug-screening process from conventional practice to one that produces innovative therapeutics; reversing the development pipeline by starting in humans first; and considering combination therapies for nervous system disorders.

Develop Public–Private Partnerships to De-Risk Research

Collaboration and the ability to share risk and costs among entities are two reasons why public–private partnerships are important for successful drug development, especially with current resource constraints. In addition to pharmaceutical companies stepping back from the drug discovery process, venture capital (VC) typically does not take risks beyond 7 years. There is a need to spread the risk among discovery, preclinical proof of concept through Phase I and Phase II trials, a participant noted. Bountra commented that industry is good at processes that require scale and infrastructure (e.g., high-throughput screening, lead optimization, manufacturing); target discovery is the main challenge. Academia and industry could work together to help develop targets that can be converted to drugs. Potter agreed and stated that the current drug development model does not allow the field to translate molecular science into new therapeutics, and the idea of sharing risk across several groups (e.g., stakeholders, advocacy groups, government, industry, and academia) would help investments from industry to focus on compounds that are more likely to succeed. Lawrence Goldstein commented that rather than trying to de-risk, the field could try to identify key bottlenecks in the system that allow risk tolerance and a degree of uncertainty.

Bountra is among a group of researchers spearheading a public–private partnership aimed at overcoming the many challenges facing drug development for nervous system disorders (Norman et al., 2011). The partnership consists of academic researchers, regulators, and several pharmaceutical companies. The goals are to de-risk drug development, increase efficiency, and accelerate new knowledge. The group hopes to achieve these goals by

• focusing on epigenetic targets;
• making novel reagents freely available;
• evaluating drugs in human primary cells;
• publishing all data immediately; and
• pushing the precompetitive boundary to Phase IIa when evaluating novel targets in patients.

Bountra elaborated that he and his colleagues are creating a public–private partnership in which resources are shared, which in turn will help de-risk the drug development process to Phase II. Bountra concluded that pioneering drug discovery can be challenging for any single organization and suggested that advances might occur through pooled funds from both the public and private sectors, improved access to reagents, and collaborations with patient groups and regulators to de-risk novel targets. After a public–private partnership advances the preclinical space forward toward Phase IIa, industry could be better situated to generate new drugs with greater accuracy and less risk, Bountra said.

Kazumi Shiosaki, managing director at MPM Capital, agreed with Kiran Reddy, principal at Third Rock Ventures, that there is a need for public–private partnerships to help alleviate the risk for neuroscience venture investments. VC firms are continuing to invest in the preclinical stages of drug development and are optimistic about emerging therapeutic approaches and technologies in the field. Shiosaki described how VC firms such as MPM work to reduce the time between their investment and a pharmaceutical company acquisition liquidity event by using various types of structured acquisitions to share risks and rewards. In a structured acquisition, large pharmaceutical companies make the bulk of their payments for the startup (known as “earn-outs”) contingent on the success of clinical trials or commercial activities. More recently, when there are drugs in development, the acquisition is made using an “option to acquire” agreement for the pharmaceutical company to acquire the startup if programs advance successfully. The pharmaceutical company pays an option fee that brings in non-dilutive capital to advance the drug project. The startup and the pharmaceutical company need to have a clear agreement and definition of the milestones that trigger the option.

Several participants suggested there is an increasing trend for lead generation and other steps in preclinical testing to be done by organizations other than drug companies (e.g., government, academia, and biotechnology companies). Potter noted that some drug companies choose to minimize risk by devoting their resources to drugs that are farther along the development pathway. Consequently, NIH has become more actively involved in supporting preclinical drug development. Through its National Center for Advancing Translational Science (NCATS), NIH supports new programs in preclinical development, one of which is the NIH molecular libraries program.¹ The program consists of a network of national laboratories whose aim is to generate novel small molecule probes by performing high-throughput screening, secondary screens, and medicinal chemistry. In addition, NIMH is supporting the Fast-Fail Trials (FAST) initiative² in the discovery of psychiatric medications by providing a rapid means for researchers to test new or repurposed compounds for their potential therapeutic use.

NCATS is working with other NIH institutes involved in neuroscience research to establish a clinical network in which experimental medicine can be performed in an efficient way, said Austin. Koroshetz reiterated that smaller and medium-sized institutes that want to participate in translational research could benefit from leveraging what is expected to be larger and more stable resources from NCATS. The initiative is designed to forge a network of centers that have harmonized information (both non-disease-specific and disease-specific). To help investigators understand FDA requirements and to facilitate communication, NCATS is engaged in conversations with FDA about having FDA staff members serve as consultants to new investigators. A few participants noted that by supporting public–private partnerships, government partners, such as NIH, might improve and accelerate drug development through a focus on understanding disease mechanisms and the underlying biology.

Extend the Precompetitive Space

In addition to public–private partnerships, collaboration in the precompetitive space is important to identify and validate novel targets while reducing the replication of studies. Bountra suggested that by extending the precompetitive boundary to Phase II, academic and government investigators would be included in research that establishes proof of clinical mechanism of a drug in humans before industry makes a huge investment. Sharing the risk and cost among all sectors could help advance drugs along the development pipeline (Bunnage, 2011). Several initiatives are under way to encourage collaboration in the precompetitive space to facilitate drug development.

Improve Reproducibility by Increasing Preclinical Study Standards and Sharing

Public–private partnerships and collaboration in the precompetitive space create opportunities to identify areas of improvement in the field. Two points that several participants highlighted were the need to increase statistical and research standards of preclinical studies and the creation of a central portal to share preclinical data. The overall end goal is to improve the reproducibility of preclinical data, which could further researchers’ understanding of what was or was not successful during preclinical trials before investing their time and resources into similar targets or drugs.