Connecting Basic Research and Health Care Needs
Important Points Highlighted by Individual Speakers
• The costs of research and health care have increased considerably, but the falling price of genomic technology provides an opportunity to use genomics as a systems optimizer to reduce medical costs.
• The efficiency of scientific translation can be improved through the use of milestone and outcome-driven management approaches, along with scaling and logistical methods, to maintain a focus on research goals and shorten the time between discovery and development.
• The application of genomics in cancer can be used as a model to demonstrate how a stratified approach can categorize tumors and guide therapy choices.
• The success of personalized medicine relies on the collective efforts of clinicians, patients, and researchers to inform research inquiry and translation.
• Speed, flexibility, teamwork, skilled management, and powerful technologies will likely be the hallmarks of the success of research in the future.
Two speakers opened the workshop with broad overviews of the pathways from basic research to improvements in health care. Edison Liu,
president and chief executive officer of The Jackson Laboratory, explored ways of improving the efficiency of translation by taking advantage of scale and logistics and by using effective management to guide research on the basis of outcomes and milestone achievements. David Huntsman, associate professor of medicine at the University of British Columbia, used cancer as an example of the potential—and complexity—of personalized medicine.
The costs of research and health care have risen, Liu said. Genomics, as a diagnostic tool, can be used as a systems optimizer, “where each diagnostic makes money by saving money for the system.” Companies have begun to think about their business models in different ways to design cost-saving solutions to health problems. Veracyte, Liu explained, develops molecular tests to examine fine needle aspirates from thyroid nodules to determine the likelihood that a tissue is benign. As a business model, this results in minimization of the need for unnecessary surgery for a complete thyroidectomy, reducing direct medical costs by more than $120 million per year (Li et al., 2011). Big science should include more efficient research processes and cost-effective outcomes, as opposed to just focusing on profit-generating ones, Liu said.
“My central premise is that our academic biomedical research enterprise is inefficient relative to the technologies that we have available today. We need a new mindset, and that mindset may be characterized by mission-oriented research,” Liu challenged. Mission-oriented research can be described as research driven by milestone accomplishments and focused on outcomes, he said. These goals can be accomplished by quality scientific managers who “execute with speed, flexibility … who can assemble functional teams quickly and disassemble them quickly … who can embrace powerful technologies and actually retire [out-of-date] technologies,” Liu said.
Liu provided examples of organizations that are using this new mindset to work together as teams to achieve common goals. The Genome Institute of Singapore, which Liu helped develop, supports those with diverse skills to engage in collective decision making around the topics chosen for research through the development of integrated platforms. The Janelia Farm Research Campus in Virginia, which opened in 2006 and is supported by the Howard Hughes Medical Institute (HHMI), is an example of an effort to provide scientific focus for long-term grand challenges in an environment which offers opportunities for cross-discipline collaboration (Waldrop, 2011). Using Bell Laboratories and other successful research models, HHMI created the Janelia Farm to combine particular areas of
research focus with individuals with diverse skills and the freedom to pursue good ideas. The model provides researchers with a fixed time commitment for their work to reinvigorate the team and continually provide fresh ideas. As an example of how successful projects are managed at milestone-based organizations, Liu said that at the Defense Advanced Research Projects Agency, “project managers are empowered, are paid well, and have significant influence in how the component parts are run.” He also cited the U.S. military’s use of a 20-year scanning horizon as an example of the importance of giving strategic attention to an issue to make progress.
Employing Scaling and Logistics
The appropriate use of scaling and logistics can also improve the efficiency of the translational pathway, Liu said. He explained that the process of translating scientific discoveries to clinical practice can be accelerated by scaling. Genomic technologies were the first examples of what he referred to as a “mature research infrastructure,” in which time-consuming technical and logistical tasks were replaced by automated or simplified ones so that more time could be spent on higher-order science. “We have never in this field seen this kind of efficiency in the tool sets that we have provided,” said Liu. “This is a whole new mindset—which, by the way, engineers and physicists have [had] for the last 100 years: the ability to scale with orders of magnitude in terms of efficiency.” For example, restriction nucleases are now readily available commercially and are inexpensive and ready to use, whereas 20 or 30 years ago, scientists needed to prepare their own. This has provided opportunities to expand science from production of a single item to production by use of the assembly line process, he said.
Logistics can also significantly shorten the length of time from discovery to clinical use of a drug because existing drugs may have uses for other indications. Many of the components in the translational pathway are already in existence, so once they are identified, it is just a matter of assembling the product. Liu used the analogy of computer parts being manufactured in different locations and then assembled together in one place to describe the increased efficiency that has started to occur in research. With improved technology and the appropriate use of scaling and logistics, a single investigator can conduct much of the work, Liu said.
In summary, Liu described that it took 41 years from the time of the genetic discovery in 1960 linking the Philadelphia chromosome as a marker for chronic myelogenous leukemia before imatinib (Gleevec) was developed and approved for patient use by the U.S. Food and Drug Administration (FDA) (Capdeville et al., 2002). In contrast, the more recent discovery of the kinase-activating gene fusion EML-ALK in a lung cancer patient resulted in the 2011 FDA approval of crizotinib (Xalkori), a process that
took just 4 years (Gandhi and Janne, 2012). The challenge now is to take advantage of a mature infrastructure and the increasing scale of the enterprise to advance the progress of basic scientific research, said Liu.
GENOMICS AND PERSONALIZED CANCER TREATMENT
Only some of what is called “basic research” is truly that, meaning that it is an interesting story but does not relate to a translational pathway, Huntsman said. True translational medicine requires that specific clinical questions and translational pathways be identified, but the current health care system is not prepared for the translational initiatives that Liu mentioned earlier, Huntsman said. “Once you start mentioning a disease and you have a focus on the disease … at that point the health care needs you are trying to address would have to be defined and a translational pathway predetermined.” However, Huntsman cautioned that the power of collective research efforts can sometimes be weakened when those conducting basic scientific research are forced to decide up front how their research will be clinically relevant, said Huntsman.
In referring to the health care system in British Columbia, Huntsman pointed out that the structure focuses on answering questions of cost-effectiveness, treatment effectiveness, and the quality of the patient and clinician experience. Without collective efforts, the delivery of equitable personalized health care in this system will be a challenge, which is a reason why cancer makes a good model for personalized medicine, as it is a disease consisting of the interplay between two genomes—that of the host and that of the tumor, Huntsman noted.
Cancer can be considered three different diseases because of its characteristics: some cancers involve many different events such that thousands of cases may need to be examined to understand the drivers involved; other cancers are more monomorphic and have single mutations driving the disease; and finally, some cancers are in between these two states and contain multiple mutated pathways, Huntsman explained. These different disease pathways necessitate the use of different approaches to translate related discoveries in basic science into the clinic. A given type of cancer may have several different subtypes for which driver pathways, prognoses, and potential treatments are distinct. These attributes of disease emphasize the importance of defining the questions to be answered when a translational research program is devised.
Cancer: A Model for a Stratified Approach
A major goal of the translation process in cancer is to move from generic management, which does not account for tumor heterogeneity,
toward more individualized or stratified cancer treatments. Successes have been achieved to date, but many of the available approaches are still too crude for use in practice. Cancer stratification can be defined by molecular features, but a critical question is how to group tumors for treatment. One approach to address this issue would be to categorize tumors into “finer and finer groups,” Huntsman said. He gave the example of a mucinous carcinoma of the ovary, which represents about 4 percent of ovarian tumors, but close to 20 percent of mucinous carcinomas have a HER2 gene amplification, for which a targeted therapy, trastuzumab, exists (Anglesio et al., 2013; McAlpine et al., 2009).
The challenge with finely stratified disease is that it makes it difficult to get enough patients to conduct clinical trials of treatments, Huntsman stated. A potential way to identify larger numbers of patients to facilitate clinical trials may be through social networking, but this has not yet been proven, Huntsman said.
Another approach used to categorize tumors, Huntsman explained, is to stratify cancer by molecular features that are shared among particular types of cancer (NRC, 2011). This stratification approach would enable the construction of a “molecular taxonomy of cancers that moves beyond subtype,” but it does not take into account three key features of cancer: heterogeneity within tumors, the activation status of mutations, and cell context, Huntsman said.
Genomic heterogeneity is particularly worrisome, Huntsman said. A one-time single sampling of a cancer may not reveal the biomarkers needed to determine treatment because the molecular signature of the tumor may change over time. A mutation critical in the early development of cancer may no longer be active, or mutations that are invariably active in one cancer type may no longer be active in another. One way to identify active mutations is to use existing bioinformatics tools to understand how mutations disrupt networks and then create lists of common and active mutations. Cell context is also important: a melanoma and a colon tumor with the same mutation may not have the same response to one targeted therapy. “We can’t just blindly move ahead with treating mutations and not cancer,” Huntsman said. Importantly, the system is lacking appropriate models that accurately reflect disease. For example, cell lines need to be carefully chosen for use in experiments to validate findings in a disease context that makes the most sense.
Many of the same questions surround the treatment of relapsed disease, Huntsman said. Treatment decisions need to be revisited over time as tumors change. Average samplings over time to detect circulating tumor mutations that are free in the plasma may be one way to approach this.
Today, there are often too many data for effective on-the-fly integration in the clinic, but better integration will be possible as genomics and
bioinformatics become common decision support tools. In particular, care communities could reformulate themselves around high-content medicine and an informatics backbone. Those that do so, especially in general practice, will embrace a future that will encourage others to do the same, Huntsman said. He said that he looks forward to a time when instead of working as a pathologist and diagnostician he will work on an integrated team with other diagnostic specialists, including informaticians, to collaboratively interpret data and make decisions.
IMPLEMENTING CHANGE FOR THE FUTURE
Liu made several projections for the future of genomic medicine. The genomes of all children with developmental disorders will be sequenced, he said, and all cancers will be sequenced. Medical analytics in a secure environment with honest brokers will be important not only to drive efficiency but also as a revenue generator for biotechnology companies. Such efforts will improve the effectiveness of delivery of health care by creating better outcomes at reduced costs. Progress will be iterative, he said, with each step forward built on the basis of experience and new knowledge.
The most impactful space to begin making changes that would improve the translational pathway would be in general practitioners’ offices, Huntsman suggested. In this way, the focus would be on the patient–caregiver interaction, and this interaction would have more of an impact on improving patient care decisions than a meeting with a specialist for treatment of, for example, an advanced-stage cancer would. Academic clinicians would be an important group to ask which challenges within the translational pathway are crucial to work on.
Liu and Huntsman both addressed where the research and clinical care communities could begin to implement systemwide changes that would have significant impacts on the translation of basic genome science into the clinic. Liu noted that hospital systems would be a good starting point, because the inefficiency of translation could be considered an issue of competitiveness in a challenging economy and innovation could be used as a cost-lowering tool. The important stakeholders for discussions of this would be the health insurers and the genomics scientists. Liu said that having access to the wealth of data from insurers on survival data from patients with subsets of cancer, for example, would be very helpful. The Jackson Laboratory, he explained, is working with Aetna to develop a model of outcomes and cost reduction for its 1,400 employees. Huntsman mentioned the importance of working with the community in new ways to improve the health of medically underserved populations. He discussed that the Ministry of Health in Canada is interested in the health of the country’s First Nation populations, whose health lags behind that of the majority population.
Time, cost, and information deficits will continue to be challenges to establishment of an improved pathway of translation. Data sharing would alleviate some of these constraints, said William Rutter, chair and founder of Synergenics and session moderator. The President’s Council of Advisors on Science and Technology Report to the President on Propelling Innovation in Drug Discovery, Development, and Evaluation would help provide a strategy for the coupling of transparency in data sharing with regulatory approval (PCAST, 2012). “Both speakers have articulated a need for systematic change, systemic reform [of the translational research pathway].… [The] confluence of information provides the basis for change, and the cost issues demand it,” Rutter summarized.
