What is the one thing that would enhance our ability to translate insights from human genetics into new medicines within the next 3 to 5 years, asked Nadeem Sarwar, the workshop chair and president of Andover Product Creation Innovation Systems at Eisai Inc. One example of a disruptive technology that could accelerate genomics-driven drug discovery is being developed by the Tissue Chip for Drug Screening program at the National Center for Advancing Translational Sciences (NCATS); it is an in vitro platform that uses human tissues to assess the safety, efficacy, and toxicity of potential drugs. Following a brief presentation on the Tissue Chip, individual workshop participants revisited Sarwar’s question and focused on two main topics: potential ways to foster a broader view of biology that fuses genetics with research on many other facets of health and disease; and ideas for sharing information, including data from genetic cohort studies, to enhance progress in developing new disease treatments. This final chapter of these proceedings synthesizes these discussions. Box 5-1 provides a list of possible next steps for genomics-enabled drug discovery efforts that were drawn from remarks made by individual speakers in the final workshop session.
As an example of a technology that could produce disruptive changes in the drug discovery and development process, Danilo Tagle,
the associate director for special initiatives at NCATS, described the Tissue Chip for Drug Screening program.1 The goal of the Tissue Chip program is to develop an in vitro platform in which human tissues are embedded on a chip to evaluate the safety, efficacy, and toxicity of promising drug candidates. Data from the Tissue Chip program could also assist regulatory agencies with making evidence-based decisions, Tagle said. As part of the program, 10 human physiological systems—circulatory, endocrine, gastrointestinal, immune, integumentary, musculoskeletal, nervous, reproductive, respiratory, and urinary—will be functionally represented by human tissue constructs. The various systems will be captured in a platform that is genetically diverse and has relevance to the physiology and the pathology of the relevant tissues. The chips will be modular and reconfigurable, tissues will be viable for at least 4 weeks, and the materials will be made widely available, Tagle said.
The development of tissues on chips requires input from cross-disciplinary teams consisting of engineers, material scientists, and cell biologists, Tagle said. These teams must take many factors into account when creating the platforms, including computational design, functional readouts, host response, innervation, bioreactors, perfusion, spatial and temporal patterning, structure, cell type, and scaffolding. To survive, the tissues require microvasculature, dedicated bioreactors optimized for the appropriate tissue response, and realistic immune responses, Tagle said. For example, researchers are currently working on a “gastrointestinal system on a chip” that can undergo peristalsis and secrete digestive enzymes. Another system seeks to re-create microvasculature with an embedded colon tumor, Tagle said. The “heart on a chip” has been used to model a rare disease called Barth syndrome and search for potential treatments (Wang et al., 2014).
Tissues on chips have the potential to incorporate many of the genetic resources discussed at the workshop and have numerous potential future applications, Tagle said. In 2016, NCATS plans to use the tissue chip technology to model rare diseases, and it will further expand the program to more common diseases in 2017. There are also plans to harness gene-editing technology to introduce various polymorphisms into induced pluripotent stem cells to look at individual drug responses. “My goal eventually would be to be able to conduct clinical trials on chips,” Tagle said, which could make preclinical and clinical research more cost effective.
Large-scale population cohorts with extensive genetic and phenotypic information are rich sources of information about potential drug targets, said Geoffrey Ginsburg of Duke University in summarizing the session on large genetic cohort design (see Chapter 2). However, he continued, findings from cohorts need to be linked to the underlying mechanistic biology in order to improve drug discovery and development. Russ Altman of Stanford University said that genetics research is an integral component of any data portfolio and can help drive drug discovery, but he cautioned that genetic data on their own are not sufficient to get the job done.
“When we understand the biology, we can make enormous inroads in drug discovery,” said Tim Rolph of Pfizer. That has happened with certain types of cancer over the past few years—with childhood leukemia, for example, for which survival rates have substantially increased. The same thing could happen with other common diseases, such as cardiovascular disease, Rolph said. Researchers should focus their efforts on areas of health care where there are unmet needs with regard to the underlying biology and try to create targeted therapies, he said. Having a clear understanding of disease pathogenesis is extremely important, Rolph said, because without that information, drug developers have to take “leaps of faith” without good ways of managing risk.
The need for a broad view of biology was made apparent in the presentations on genome-enabled discovery activities, said John Carulli, the director of precision medicine at Biogen and moderator of the session on current genome-enabled discovery activities (see Chapter 3). Identifying genetic information to help facilitate the drug discovery process is only the first step, he said. The molecular, cellular, and organismal consequences of target modulation—including the effects of such modulation on the health trajectory of patients—are also critical to success in drug development. In-depth biological information is essential, Carulli said, adding, “We need rich phenotypic data, and we need it to be longitudinal.”
Polypharmacology, or the design or use of pharmaceutical agents that act on multiple targets, has the potential to increase our understanding of molecular disease mechanisms. It is much easier to treat a disease when you can take multiple shots at a pathway, said Russ Altman of Stanford University. Polypharmacology is becoming increasingly common in medicine, he noted. “We treat HIV with three drugs, we treat tuberculo-
sis with four drugs, we treat depression with two to four. . . . The targetability of a pathway statistically is better than the targetability of a single element of the pathway.”
Gathering rich and accurate phenotypic data can be more complicated than gathering genetic data. For example, one workshop participant commented that while the concept of 23andMe is exciting, the population that the company is studying may not accurately represent the general population. And while disease registries tend to have relatively good phenotypic datasets, electronic health records generally do not contain research-grade phenotypic data, the participant noted, saying “phenotypic data has not yet reached a level that we’ve achieved for genotypic data.”
While phenotype measurements can be subjective, some fields are beginning to rethink the boundaries of phenotypic descriptions so that they are more biologically meaningful, said Mark Daly of the Broad Institute of the Massachusetts Institute of Technology and Harvard University. Daly noted marked progress in the area of mental health: “There’s now a push toward developing simple-to-apply [and] broad tools that can characterize mental state in a way that will be able to be consistently applied across populations around the world.”
Phenotypic descriptions can be captured at various levels of detail, Altman noted, ranging from relatively simple measures such as the presence or absence of a condition, to more complex phenotypes such as the status of multiple biopsies. Informaticists are currently developing ways of using proxy variables, algorithms, and post-processing to extract valuable phenotypic data from the information that is available, he said, adding that there are ways to be very clever in extracting the data needed to answer a particular scientific question.
In many cases, callbacks to cohort participants for deeper phenotyping will be necessary, Ginsburg said. For example, there are many mutations known to exist in the population for which there is no knowledge about the phenotypes associated with them. One possible approach would be to use information gathered through unconventional means to inform genetic studies. For instance, Ginsburg suggested, perhaps it would be possible to use the 23andMe survey platform more ubiquitously throughout cohort studies. Another possibility would be to use mobile
health technology as a way to capture phenotypic information in a standardized way, he said.
The broader view of biology that is developing could accelerate genomics-enabled drug discovery and should also incorporate the perspectives of patients, said one workshop participant. Genomics-enabled drug discovery is not moving as fast as it needs to move from the perspective of patients, the participant said, even though many patients are willing to share their genetic and medical information, which could result in the discovery of a wealth of potential drug targets.
A widespread cultural shift toward incorporating the needs and desires of patients during the design of research is something that could enhance outcomes, said Lynn Matrisian of the Pancreatic Cancer Action Network. Such a shift will create a sense of urgency and lead to the acknowledgment that one person cannot do this, that it is going to take working together as a team, she said. This approach is becoming more common, she added, but it is not yet widespread. New incentives, structures, and goals will be able to facilitate a cultural shift in this direction, she said.
On the subject of the involvement of patients, Lon Cardon of GSK expressed concern about the way that hope can sometimes turn into hype. It is difficult to predict what will happen in the next 10 years, he said, but overpromising and underdelivering remains a problem. However, hype tends to be cyclical in the field of genomics, he added, and the field is in a less hyperbolic phase now, because new drugs are just starting to emerge from research.
The potential for collaboration and data sharing to enhance drug discovery has been demonstrated in the past, Mark Daly said. Over the past 7 years, more than 100 genetic associations have been discovered for schizophrenia, he said, which happened because more than 50 independent groups around the world were willing to deposit their genetic data on a single server and engage in a collaborative analysis activity
(Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014). It is likely that these advances would not have been achieved in such a short amount of time if the participating groups were not willing to work collaboratively, he said.
The opportunity for data-sharing success, as exhibited by collaborations in the mental health field, underscores the need for new collaborative models that would discourage investigators in academia and industry from keeping their data to themselves, Daly said. Genome-wide association studies provide an opportunity to share results in a way that addresses a more nuanced set of questions that go beyond simple genetic associations and attend to details about the tissues, cells, and developmental time periods in which genetic variants are active, Daly said.
Data Sharing as a Way to Reduce Research Duplication
The sharing of genetic data is a powerful opportunity because it engages an entire community of data analysts, each of whom might have been previously working in individual data silos, Daly said. Geneticists need to do a better job of reaching out to the broader molecular biology community with well-annotated genetic results in order to make these data more accessible to the entire research community, he continued. Today, a great deal of redundant research activity takes place as people do the same set of basic data processing and analysis activities. Currently, the discussion is too focused around raw data access, Daly said, arguing that there should instead be a focus on collaboratively completing the tedious elements so that the entire community can move on with the research.
Other speakers also pointed to the duplication that could be eliminated through expanded data sharing. For example, Chas Bountra of Oxford University observed that many companies work on the same drug target, and if a target fails, considerable time and resources can be lost. While replicating research results is essential, time and money could be saved by reducing the amount of unnecessary experimental duplication, particularly before a target is validated.
Factors That Could Contribute to Expanded Data Sharing
A clearer and more widely accepted definition of the boundaries of precompetitive research could enhance information sharing, as would standards to increase interoperability, Cardon said. “The competitive
part is making the drugs, but if we can share resources on unraveling the function of these targets, then maybe we can do a better job of developing drugs,” he said. Even companies like 23andMe share data extensively. 23andMe has many academic collaborations, Scheller said, even though the data cannot leave the company because of its informed consent process.
Both the benefits and risks of information sharing should be openly communicated in order to optimize partnerships and drug discovery, said Rajesh Ranganathan, who at the time of the workshop was the vice president of science and regulatory advocacy at PhRMA. It is important to figure out what works for all parties involved, Ranganathan said. In several successful collaborations, the parties have determined what each can bring to the table and how they can potentially benefit. Leadership is also important in bringing disparate groups together in collaborations, Ranganathan said. As a result, collaborations will vary in their degree of success, depending on the type of leadership involved.
To increase data sharing, Altman suggested establishing a website, potentially named geneticcohorts.gov that would function like ClinicalTrials.gov, where all applicable clinical trials in the United States must be preregistered before beginning participant enrollment. Such a website could announce the existence of a genetics-based cohort study, its size, the types of metadata associated with it, and information on whom to contact to initiate a collaboration. The site would not contain the raw genetic data, but inclusion in the site could be a condition for publication, he said.
Trust between patients and researchers is essential in data sharing, Ginsburg said, which means that “our job is to educate.” Most of the organizations conducting population studies are reaching out to communities and creating principles of engagement, but studies need to treat these individuals of the community as partners, not just as subjects, he said. “Shaking the hands of the participants and having them understand the transaction that’s going to occur and how they are going to benefit from it is something that we have to take quite seriously.”
Sharing appears to be common in some areas and organizations but not everywhere, Cardon said, adding, however, that the tide may be turning. From the business perspective, sectors have become increasingly entrepreneurial and collaborative, creating alternative ways to think about drug discovery and development. “We have an opportunity that is not just once in a generation but once in a lifetime,” Cardon said. “The information sitting in front of us has the power and the potential to
transform the way we think about the next generation of medicines.” The technologies and data exist or are imminent. But an acceleration of progress will require working together to translate research into medicines, which in turn will require innovative and not just incremental thinking. “It’s really exciting, but it’s incumbent upon us to make it happen,” Cardon said.
Three themes were touched on by individual speakers during the workshop: hope, incentivizing entrepreneurs, and patient-centricity. First, it is critical to broadly communicate the opportunities that genomics-driven drug discovery provides, Sarwar said. One example of an exciting genomics-enabled prospect is the ability to understand which subsets of patients are most likely to benefit from a particular medicine, he said, and this idea has changed how new medicines are developed. The realization that genetic support for drug targets increases the odds of success has underlined the importance of genomic research. This new perspective has contributed to substantial advances in the ability to treat and potentially cure diseases, and hope for continued advances is a strong motivator and enabler. Agreeing with Cardon, Sarwar said, “This may be a once-in-a-lifetime opportunity for us to truly change our odds of success for developing new medicines.”
Second, scientific and commercial advances can also act to incentivize entrepreneurs. The challenge, Sarwar said, will be to put in place the structures to support entrepreneurship. Would better ways of integrating and applying data unlock new abilities to make medicines? Can the right combinations of expertise be brought together to begin realizing those opportunities? “Are there business opportunities that we’ve missed?” he asked.
Finally, patient-centricity needs to remain at the forefront of drug discovery and development, Sarwar observed. Patients are the primary drivers of research and drug development. What endpoints address patient needs? What are measures of progress versus measures of impact? What medicines do patients want, and how can these desires be met? “We don’t make medicines because it’s easy, or because it’s a guaranteed return on investment,” Sarwar said. “We make medicines because they are needed by patients. It’s a privilege to be able to make new medicines, and it’s a privilege to try to understand why diseases are caused.”
Each sector involved in drug discovery and development is now in a position to help realize the potential behind that privilege.