Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 3
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary 2 Translation of Innovations A BROAD PERSPECTIVE Robert M. Califf, M.D., MACC Duke Translational Medicine Institute and Duke University Medical Center Biomedical science is advancing at an amazing rate, yet the translation of that science into better health outcomes has not kept pace. Much of this lag is due to non-technological reasons, including financing, regulation, and cultural issues. Another factor is that the rewards for researchers who promote innovation are increasingly disconnected with the healthcare needs of society at large. Translation is a fragmented and uncoordinated process that, with few exceptions, takes 25 to 30 years from initial scientific discovery to the delivery of a therapy to the people who benefit most (Figure 2-1). While basic discoveries occur predominantly in academic medical centers funded by the National Institutes of Health (NIH), the process of translating these discoveries almost always begins in the medical products industry, where a basic discovery is followed up with a period of specifically directed preclinical activity intended to test whether the putative therapeutic target is indeed viable. The next step is determined by a decision-making process that comprises multiple steps and includes assessments that link financial support with the probability of success; if the decision is to move forward, then the next stage of development is undertaken by clinical research orga-
OCR for page 4
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary FIGURE 2-1 Translation of innovations. SOURCE: Califf, 2007.
OCR for page 5
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary nizations from the medical products industry, contract research organizations, or academia. The early period of human subjects research, commonly called “proof of concept” or phase I/IIa, is characterized by the introduction of novel therapies into either healthy volunteers or a carefully selected group of patients; if there are no red flags, this work is followed by a comprehensive set of clinical studies, known as phase III trials. Data from these phase III trials are then used by the U.S. Food and Drug Administration (FDA) and other national and international regulatory bodies to make decisions—based on criteria that vary depending on which division of the FDA is involved or which country is doing the evaluation—about whether the therapy is ready to be introduced into clinical use. After a therapy is approved, it is supposed to reach the appropriate people in the approved manner through a competitive system that includes health systems, hospitals, clinical practices, purchasers, and sales representatives for the product or technology. Ultimately, when the therapy’s patent protection expires, its price will diminish, and the health of the entire community will benefit from the wider access thus afforded. This system has generally worked well up to now, as evidenced by the steady decline in mortality in the United States since 1900, a decline only briefly interrupted by the 1918 flu pandemic. And while much of the decline during the first half of the 20th century was due to clean water, sewers, antibiotics, and better nutrition leading to a reduction in mortality from infectious diseases, a significant proportion of the decline since then has been attributable to advances in treatment, with the prevention of infant mortality and the treatment and prevention of cardiovascular disease playing the largest role. Despite these achievements, however, key issues must still be addressed concerning the translation of scientific innovations into effective therapeutics. We now have information systems capable of providing detailed data on leading causes of death and disability, for example, and these data show that the benefits of technological advances have not been evenly distributed (Figure 2-2). Such information can be helpful in identifying new directions in which to focus the efforts of the translational enterprise. Challenges Facing Translational Medicine Our current general scheme of focusing on discovery science in academic centers and trusting for-profit industry to handle the diffusion of technology continues to be the most sensible path to follow. But along that path are major hurdles that must be cleared, particularly at the translational interfaces between discovery and commercialization and between commercialization and public health. In the arena of drugs and biologics, for
OCR for page 6
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary FIGURE 2-2 Life expectancy at birth. SOURCE: Adapted from Harper et al., 2007. instance, although novel targets afford bigger potential returns on investment, investors often shy away from them because of the risks entailed. Pursuing an already-proven target gives a much higher probability of success, which causes “follow-ons” to be seen as a better bet on average and leads investors to often—and understandably—choose the safer option. The net effect of these considerations is a risk-averse industry that pursues fewer novel, innovative pathways. In the arena of genomics-based diagnostic testing and therapeutic decision making, for instance, the intersection of diagnostic testing and therapeutics is plagued with regulatory ambiguity, and the prospects for reimbursement are unsure. Such uncertainty directly affects willingness to invest. In terms of health services, enormous investment will be required to change current practices. Forces that encourage change in health care services (i.e., the Internet, consumerism, information technology, -omics,1 1 -omics refers to a biological field of study that ends in the suffix omics, for example, genomics, proteomics, metabolomics.
OCR for page 7
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary medical technology, and Congress) are offset by countervailing pressures (i.e., regulation, financing, a fragmented marketplace, professional autonomy, and, once again, Congress). Many observers believe that these forces have created an equilibrium that discourages innovation, but there is no consensus about how that equilibrium can be changed while still maintaining the fundamental safety net created by the regulation of technologies through objective, empirical assessment of the balance of risk and benefit. The high cost of developing a new product is one example of the difficulties facing innovation. A study conducted in 2003 by DiMasi and colleagues found that research and development costs for a new drug in the United States averaged a total of $800 million in 2000 dollars, up sharply from the estimated $231 million that such research and development cost in 1987 (in 1987 dollars) (DiMasi et al., 2003). The most recent published data provide an astonishing estimate of $1.4 billion per successfully developed drug. An important component of this figure is the cost of capital during the protracted period of drug development. Unfortunately, the U.S. clinical research system is increasingly recognized as a bottleneck in the process of therapeutic development, as clinical research takes longer and is measurably more expensive to accomplish in the United States than in other countries, while the quality of the research itself may be inferior to that conducted in other parts of the globe. Furthermore, the application of therapies in the United States is measurably inefficient—not only are the costs of the therapies much higher here than in other countries, but the therapies have inferior results in terms of longevity and functionality of the population. Another potential deterrent to innovation exists at the level of practice. The movement toward evidence-based medicine has pushed practitioners to have evidence for what they are doing. On balance this is clearly a favorable development. It gives patients and consumers much more confidence that the treatments they receive are appropriate to their needs and that they are administered correctly. The demand for evidence, however, can have a stultifying effect on innovation if it is employed ineffectively and without the application of modern methods and scientific insight. Incentives should be developed to foster innovation. The current U.S. health care system has many incentives to seek efficiency in the delivery of technologically sophisticated, expensive approaches for those who can afford them. There is a great disincentive, however, to providing low-cost, efficient health care to the people who are experiencing most of the death and disability in the United States. Despite astounding advances in biology, ensuring that innovations reach those members of society who stand to benefit most from them—and thus that these innovations will have the largest possible effect on the rates of death and disability—is proving especially difficult.
OCR for page 8
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary Overcoming Translational Blocks Along the translational pathway there are several blocks that slow progress from the identification of a potential biological system that could be attacked as a target to the translation of that concept into the first human studies. First, the high levels of risk involved in the process limit investment interest. Second, there is a large gap between scientific advances and the regulatory science needed to predict and evaluate product performance. Third, decision making is dominated by anecdote and intuition. In order to make a prediction about the success of a possible therapy, one must know what has succeeded and what has failed in the past and then use that information to understand the probabilities of success or failure in general. If only successful efforts are made public, however, there is little basis for understanding and determining which general approaches lead to greater success and thus for figuring out where to invest efforts and funding. The Critical Path Initiative This lack of data about the factors that underlie the success or failure of development efforts is a major motivating factor for the FDA’s Critical Path Initiative,2 which aims to create a “safe haven” for sharing knowledge that can accelerate translation while at the same time doing nothing to impair the drive for competitive advantage that stimulates creativity in our system. The concepts of pre-competitive and pro-competitive spaces are key to understanding the strategy underlying the Critical Path Initiative. Generally speaking, pre-competitive knowledge advances a field as a whole before the point at which competition based on proprietary knowledge comes into play. An example of pre-competitive knowledge would be general knowledge about the operating characteristics of standard tests for pre-clinical toxicity required by the FDA. Currently, little is known about the true predictive value of these tests because abandoned projects are rarely discussed and almost never published, leaving an incomplete database of test results that renders any calculations about the value of the tests meaningless. The pro-competitive space is characterized by mutual efforts toward development of new knowledge that in the past would have been proprietary but that, through collaboration, confers an equal advantage to all interests. An example would be a generally known biomarker that everyone 2 “The Critical Path Initiative is FDA’s effort to stimulate and facilitate a national effort to modernize the scientific process through which a potential human drug, biological product, or medical device is transformed from a discovery or ‘proof of concept’ into a medical product” (FDA, 2006).
OCR for page 9
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary can use. Individual companies usually do not have enough biological and clinical data to validate a biomarker, but a consortium of companies and academic institutions may be able to do so. Companies that make best use of publicly available information about the biomarker in developing therapeutics would be the ones to receive an advantage. Continuing on the translation pathway illustrated in Figure 2-1, the next step is early-phase human studies. Many discoveries fail at this stage because of unanticipated off-target effects that are only detected in much later phase testing. A major recent example was the case of torcetrapib, a drug developed to treat abnormally low HDL cholesterol and prevent cardiovascular disease (Nissen et al., 2007). Torcetrapib failed in phase III trials, perhaps because of an unrecognized and completely unanticipated aldosterone-producing effect. To identify these types of off-target effects before they cause harm to participants in large-scale clinical trials, it will be necessary to study human systems biology in greater detail. The traditional approach to early-phase human subjects research used in the pharmaceutical industry today (measuring pharmacokinetics, pharmacodynamics, and adverse events) does not address this problem, and a new approach that uses experimental medicine units capable of highly detailed systems measurement in human subjects is needed. Researchers will need to use modern technologies, such as gene expression analysis, proteomic and metabolomic profiling, and functional imaging, to study integrated physiology more effectively. Once early-phase human studies have been conducted, research efforts move to the larger clinical trials. There seems to be a general assumption that we know how to conduct these clinical trials effectively. To the contrary, clinical trials are too expensive, too slow, and too often of doubtful quality. In fact, there are no standard definitions of quality for different types of trials (Baigent et al., 2008). Five years ago, a typical phase III trial in cardiovascular disease cost about $80 million to $140 million (Eisenstein et al., 2005, 2008). Currently many trials cost $300 million to $400 million, or even more. Such exorbitant costs become an inhibiting factor for therapeutic areas that require definitive data as a precondition to marketing. The FDA Critical Path Initiative is seeking to transform the clinical research enterprise through the Clinical Trials Transformation Initiative. The goals of this project are to enhance knowledge and standards that improve the quality of clinical trials while eliminating practices that increase costs but provide no value in return (CTTI, 2007). Key players in these efforts include the FDA, industry, academia, patient advocates, and non-academic clinical research professionals.
OCR for page 10
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary Post-Marketing Research Once a product has been approved for marketing and is released into the marketplace, it is still necessary to generate substantial additional evidence about the balance of safety and effectiveness in the post-marketing phase. Unfortunately, there is almost no money to support such research, which has the primary goal of improving the public health. Most funding for post-marketing studies comes from the company that markets the product, and most such trials are designed to expand the market for the product and thus to bolster its expected financial value to the company. Indeed, the decision about which studies to conduct is usually based on net present-value calculations, and a trial’s sponsor will approve funding only if there is a high pre-test probability that the trial will lead to a desirable result. While these studies may give honest answers to the questions asked, the questions about translation that get asked under the current system are not the ones that would be asked if the welfare of the general public were the major concern. The Reagan-Udall Foundation, which was recently created as part of FDA renewal legislation, offers a public-private partnership to provide a venue in which such public-focused studies can be designed, but political maneuvering has so far blocked funding for this effort. The endpoint of the translation pathway illustrated in Figure 2-1 is public and global health. There is a growing convergence between national healthcare issues and global ones. As is the case in the United States, financial incentives in many other countries emphasize practices that focus on expensive technology that benefits “paying customers,” while incentives to provide basic health services receive less emphasis even as the understanding of ways to meet those basic health needs improves. In Durham, North Carolina, with funding from the NIH’s Clinical and Translational Science Awards, a study is underway whose goal is to develop a deeper understanding of the issues surrounding the delivery of basic health needs. Under the current reimbursement system, there is a tension between financial considerations and public health consideration in decisions about where to locate health clinics. In particular, the sites that are likely to result in profitable practices are not where the clinics would be located if the goal was to improve the overall health of Durham County, given that the greatest burden of death and disability is located in neighborhoods in which reimbursement is most adverse for provision of health-related services. Plans are now underway to harness geospatial-temporal mapping (Miranda et al., 2005), embedded personal health records, disease registries, and wireless monitoring capabilities in order to deploy low-cost technologies capable of delivering better, more affordable health care to the people who need it most. Providing incentives to develop technology aimed at serving people and neighborhoods with the greatest burden of premature
OCR for page 11
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary death and disability would result in an enormous redirection of innovative efforts. Indeed, the New York Times recently reported that the disparities in health outcomes as a function of education and income are widening rather than narrowing in the United States (Pear, 2008). Positive Change: The Pediatric Exclusivity Program Change, however, is possible, and the Pediatric Exclusivity Program provides a heartening example of how incentives for change can be created (FDA, 2005). In the 1990s the pediatric community became increasingly aware that many therapies used in children had no empirical data establishing their safety and effectiveness. The problem had its roots in a general sense that clinical trials in children were too risky; this community view in turn reinforced the reluctance of drug and device companies to engage in such trials. However, a determined coalition worked together to create legislation granting patent extensions to companies that agreed to evaluate their technologies by performing appropriate trials in children. Since this program began in 1997, there has been a substantial increase in drug research for pediatric indications in addition to 138 labeling changes. Li and colleagues performed a meta-analytical study aimed at quantifying the economic return to industry for 6 months of pediatric exclusivity (Li et al., 2007). Nine drugs were studied, and net economic return and net–return-to-cost ratios were calculated. The median cost per written request3 was $12.34 million. Net economic returns (minus $8.9 million to $507.9 million) and net minus return-to-cost ratios (minus 0.68 to 73.63) were highly variable, but, on balance, the net economic return to industry was favorable. Benjamin and colleagues performed a meta-analysis of clinical trials completed for pediatric exclusivity in order to quantify the dissemination of study results (Benjamin et al., 2006). They evaluated 253 studies submitted to the FDA from 1998 to 2004. Of these, only 113 were published, and efficacy studies and trials that resulted in desirable labeling changes were most likely to be published. Unfortunately, a number of the negative findings received little or no attention in the pediatric community. Nonetheless, these studies represent a positive development. Prior to this program, many in the research community asserted that clinical trials in children were not practicable. Once the incentive was put in place, however, trials were indeed undertaken. 3 The FDA issues the written request to the company. The written request describes in detail the studies needed to be eligible for pediatric exclusivity and the time frame for completion of those studies. A written request contains the indication, number of studies, sample sizes, and trials design required for eligibility.
OCR for page 12
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary Califf concluded by asking, if these incentives work, how can we deploy them in order to achieve the goals most crucial to the broad and equitable diffusion of biomedical innovations in society? UNDERSTANDING TYPES OF INNOVATION AND IMPLICATIONS FOR POLICY Kevin Schulman, M.D. Duke University According to current estimates, by 2030 about 52 percent of the entire federal budget will be required to fund the Social Security and Medicare programs. Given the annual cash flow deficits in both Social Security and Medicare, these programs will be underfunded by $2 trillion by 2030. By mid-century underfunding will reach $7 trillion (Rettenmaier and Saving, 2004). The public policy debate has not yet faced the fact that there is not enough money to support these programs; yet this is a situation that needs to be discussed and debated soon. How should one approach the issues of deficits and underinvestment in these programs? In the mid-1990s there was a crisis concerning the escalating health costs for Medicare. It was believed that the easiest way to fix the problem was to freeze Medicare spending, so the Balanced Budget Act was implemented. This was effective through the three years of the act, but, as shown in Figure 2-3, once the restrictions were removed, spending continued to increase, and the slope of that increase was steeper than before the act had been implemented (CMS, 2007). The market response to the policy of freezing expenditures was an unexpected acceleration in the costs of the program once controls were removed. If controls did not work in the late 1990s, they most likely will not solve today’s cost issues, will not solve quality issues, and will likely make things worse. So what course can be pursued? Clay Christensen examined the role of innovation in the computer disk-drive industry and put forth some ideas that have relevance to a discussion on innovation in health care. He describes the process of innovation as resulting from entry of new firms and new business models in the marketplace. In his analysis, entry results from opportunities created when products outstrip the needs of the majority of the marketplace. There is a distribution of demand by consumers for any new technology or innovation, and this distribution can be thought of as following a normal distribution. At the leading edge are early adopters, most of the popula-
OCR for page 13
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary FIGURE 2-3 Policy response: A budget freeze. SOURCE: Adapted from CMS, 2007. tion is in the middle, and there are some who are late adopters. The early adopters are of great interest to large firms since they have high demand for technology and innovation and are thought to be relatively price insensitive, Schulman said. Furthermore, the capacity of the majority of the market to use new technology increases at a much slower rate than the capacity of the early adopters. For a technology company the high demand group is a great part of the market to satisfy because, if the company can develop products or services to meet the demands of this group, their products or services can be very profitable. As firms evolve their products to meet the demands of this specific subset of the population, however, an interesting phenomenon occurs. As a result of meeting the needs of the early adopters, the technology develops in such a way as to outperform the requirements of the majority of the market. In this situation, there is a gap between the performance of the existing technology and the needs of the majority of the market. This gap creates the opportunity for new firms to bring new products to the market that might be more limited in scope than the existing technology but might be a better match on price and quality for an important part of the market. Over time these new companies actually begin to meet the needs of the general population. These two types of firms move ahead through two different types of advances. The first type of firm, the original innovator firm, moves forward
OCR for page 14
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary through sustained technology improvement. This type of innovation is called sustaining innovation. The second type of firm, which creates a new product and enters the market established by the first type, is called the “disruptive innovator.” Joseph Schumpeter wrote about this phenomenon in 1911, saying, “… as a rule the new does not grow out of the old but appears alongside of it and eliminates it competitively …” (Schumpeter, 1911). The net result of this process of innovation is the creation of higher quality, lower cost products over time. While this is generally accomplished through the entry to the market of new firms with new business models, originator firms can respond to these threats. Technology innovation is a fundamental part of the market. In fact, in most markets, technology and organizational innovation drive cost and quality improvement. How can these concepts be applied to health care? One of the things that distinguishes health care from, for example, the disk-drive industry that Christensen studied, is that the health care industry is regulated, with different aspects of it regulated to different degrees. A sustaining innovator in health care is above the regulatory barrier, that is, it has met the regulatory requirements. By contrast, the disruptive innovator that would like to enter the marketplace is below the regulatory barrier. Therefore, while the space for the disruptive innovator to enter the market is theoretically available, the regulations can deter entry. Imagine, for example, a new player wanting to enter the highly regulated hospital market and compete with Duke Hospital, which is very profitable and also one of the most expensive hospitals in the country. A new competitor to the field would have to have billions of dollars to become an innovative competitor to Duke Hospital. In reality, therefore, the space for a disruptive innovator does not exist. The administrative barriers as well as the regulatory barriers effectively bar disruptive innovation. Not all types of innovation are of equal interest from a policy perspective. From that perspective, there is a strong desire for innovation, but there is a willingness to pay a premium only for those innovations with the potential to be disruptive innovation. Since the policy goal is to improve quality and reduce costs, an implicit policy goal should be to encourage disruptive innovation and market entry to achieve this goal. In practice, however, current medical reimbursement strategies reward the sustaining innovators with premiums, making it potentially very difficult for disruptive innovators to enter the health care market. Of course, one difficulty in encouraging disruptive innovation is that it is hard to determine in advance which technologies have the potential to become disruptive innovators. There is an urgent need to better understand the relationship between incentives and market entry in order to foster technology innovation. To determine where to place incentives, one must first decide what innovations
OCR for page 15
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary are. Is molecular structure an innovation? Is the mechanism of action an innovation? What about the mode of delivery of an innovation? Is the fact that -omics is involved in some way an innovation? Is therapy that alters a treatment plan an innovation? It is important to answer these questions since the answers will shape the types of technology that are brought to the marketplace. Nexium, a product used to treat heartburn and acid reflux disease, is one of the largest-selling drugs in the United States, with sales of over $5 billion in 2006 (Astrazeneca, 2006). It is an isomer of a previous product. It is an example of what Califf referred to earlier as a follow-on product. Is it a disruptive or sustaining innovation? Should this determination enter into price negotiations? Another aspect of innovation that needs to be explored is the relationship between organizational and service innovation. What is service innovation, and can technology be a platform for that? There are several types of organizational innovation. These are generally firm responses to competitive threats from market entry and this process is how originator firms respond to disruptive innovation. Firms can respond at several levels to new product or service creation, and some can adapt their business model to a new market environment over time. Many types of organizational innovation involve the development or acquisition of new business models. A good example of innovation in an internal exploratory environment is the Lockheed Skunk Works4—the place where many of the firm’s new innovations and plans come from. Corporate venture-capital companies make investments in small firms to acquire insights into new business models. They can also make acquisitions, especially exploratory acquisitions, to acquire new products or services. This process also involves divesting older models and older technologies. Interestingly, in terms of regulatory barriers, health information technology is one of the few areas in health care where there are not yet any regulatory barriers to disruptive innovation. Schulman concluded by saying that the cost and quality pressures in health care are enormous and increasing. The easiest response is to freeze the system and lock in the status quo; the result of any such action is likely to be disappointing since this prevents organizational innovation in the 4 “Skunk Works refers to both the division of the same name within the Lockheed Martin corporation and the organizational model popularized by that division’s success at managing time-sensitive, complex projects. The latter sense is used in engineering and technical fields to describe a group within an organization given a high degree of autonomy and unhampered by bureaucracy, tasked with working on advanced or secret projects. The term is also used analogously in other fields, especially business, to describe any self-contained, semi-autonomous work-group or committee that directly manages its own projects” (http://en.wikipedia.org/wiki/Skunk_works, accessed January 18, 2008).
OCR for page 16
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary market. What is needed is a better understanding of the role of technology and of organizational innovation in the broader economy and especially in health care. If certain types of innovation can provide a solution to problems of cost and quality, then they should be part of the policy debate. If certain types of innovation can provide a solution, especially -omics, these efforts must be supported with strong market and policy messages. As stated earlier, there is an urgent need to better understand the relationship between incentives and market entry in order to foster technology innovation. LESSONS FOR GENOMICS FROM OTHER TECHNOLOGIES Annetine Gelijns, Ph.D.5 Columbia University Advances in genetics have led to a remarkably improved understanding of the genetic and molecular basis of disease, and these advances are increasingly leading to the development of new interventions in such areas as genetic testing, gene-based therapy, and pharmacogenomics. These advances permeate life and even art (e.g., the catalogue of the Museum of Modern Art offers a framed print of one’s own DNA). Advances in genetics also highlight the importance of the diffusion of innovations as well as the issue of how best to manage the challenges inherent in adopting and using genomic interventions. Research into technological diffusion finds that diffusion typically follows an S-shaped course, with adoption proceeding slowly at first, then accelerating, and then slowing down as the saturation point is reached (Griliches, 1957). There are several factors that affect the speed at which diffusion occurs. The first of these factors is the characteristics of the technology itself. These characteristics include such things as available alternatives, the marginal benefits that the new technology offers, the severity and prevalence of the target illness, and the costs and complexity of adopting the technology. This last characteristic is a particularly important consideration for genomics technologies. A second factor affecting the speed of diffusion is the collection of regulatory agencies and payers; these have become increasingly important 5 This presentation was developed collaboratively by Annetine Gelijns, Ph.D., Alejandra Guerchicoff, Ph.D., Deborah D. Ascheim, M.D., Lawrence D. Brown, Ph.D., and Alan J. Moskowitz, M.D.
OCR for page 17
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary gatekeepers of the diffusion process in health care. A third factor is the characteristics and interests of potential adopters. For some health care technologies, physicians are the sole decision makers regarding adoption. For other technologies—liver transplant programs, for example—hospital administrators and boards of trustees become involved in the decision-making process. Finally, various economic, sociocultural, and ethical factors powerfully shape the diffusion process as well. In the diffusion literature a technology is generally perceived as being static or constant; however, innovations continue to evolve as they enter clinical practice. As a result, decisions about adopting a technology are made in the face of considerable uncertainty about indications, populations, risks, and effectiveness. In recent years, the stakeholders—for examples, the FDA, payers, physicians, or patients—have sought more rigorous evidence to help guide adoption decisions. Each stakeholder brings its own distinct perspective to decisions that have major implications for quality, cost, and fairness, a fact that highlights the importance of understanding the preferences and the values of stakeholders. It is only after a new technology is put into actual use in clinical practice that there can be significant downstream learning and innovation. Such learning and innovation falls into three broad categories. First, after a new technology is put into practice, the medical profession typically refines the patient selection criteria within a given disease category. Coronary artery bypass graft surgery (CABG) is a case in point. Only four percent of patients treated with CABG a decade after its introduction would have met the eligibility criteria of the trials that established its initial value. These initial trials excluded the elderly, women, and patients with a range of comorbidities, all of whom are recipients of CABG today. Second, the process of post-marketing innovation also includes the discovery of totally new and often unexpected indications for use. The history of pharmaceutical innovation is replete with such discoveries, such as happened, for instance, with alpha blockers. These were first introduced for hypertension, but 20 years later they are an important agent for the treatment of benign prostatic hypertrophy. The discovery of such new indications of use is an important public health and economic phenomenon and accounts, for example, for nearly half of the overall market for blockbuster drugs. Gelijns commented that it will be interesting to see how the introduction of pharmacogenomics might affect this dynamic. The third type of downstream learning is the way that physicians gain knowledge about integrating a technology into the overall management of their patients. For example, the left ventricular assist device (LVAD) was approved by the FDA and the Centers for Medicare and Medicaid Services (CMS) in 2003 for end-stage heart-failure patients who were ineligible for
OCR for page 18
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary cardiac transplantation. After the device was approved, clinicians modified the operating technique in various ways. They discovered new ways to prevent infections, and they made changes in anticoagulation regimens. These changes led to a reduction in the adverse-event profile and a 25 percent reduction in the length of hospital stay. These various types of post-marketing learning and innovation take place not only with therapeutic technologies but also with diagnostic technologies. Diagnostic technologies can be used to identify abnormalities, but uncertainty remains concerning how much they can be used to infer prognoses or the need for treatment. Several controversial examples include mammography and ductal carcinoma in situ (DCIS), prostate specific antigen (PSA) testing, and magnetic resonance imaging (MRI) evidence about unbled brain aneurisms. The uncertainties have resulted in significant variations in rates of further diagnostic testing and in treatment patterns, both nationally and internationally. Genomic interventions may produce diagnostic technologies that enhance prognostic abilities. In the case of breast cancer, for example, many women receive adjuvant chemotherapy to prevent recurrence. Gene chips may identify women who have a high likelihood of developing such a recurrence and thereby allow targeting of such therapy more judiciously. But these technologies will also bring with them their own uncertainties—a positive test will not always indicate the development of disease, for instance, because a number of factors can also play a role, including variable expressivity, and environmental factors. When new technologies are introduced into health care they may be relatively primitive, which accounts for some of the slowness of their diffusion. Actual use, however, produces downstream learning, which may lead to modifications in the technology itself or refinements of its application. One such refinement, for example, is better prognostic understanding of a genomic test as a better understanding develops about the interactions among genes and between genes and environmental factors (Burke and Psaty, 2007). Additionally, physicians become more knowledgeable about how to integrate these technologies with appropriate surveillance and treatment regimens for the whole spectrum of at-risk patients (Burke and Zimmern, 2004). The clinical utility6 of such tests, however, will need to be confirmed in pragmatic clinical trials involving large, well-defined populations. Evidence is a critical factor in the diffusion of technology. The FDA plays a key role in shaping the evaluation and adoption of technology in other fields, and the agency has taken a proactive role in the area of genom- 6 Clinical utility is the degree to which a test alters medical management in a way that results in a net health benefit to the patients (IOM, 2005).
OCR for page 19
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary ics. Traditionally the FDA has regulated only those diagnostics that were marketed as kits, and CMS has had oversight over those diagnostics marketed to laboratories, but with DNA chip technology, such as the amplichip CYP 450, the FDA decided that genetic tests required a higher level of review. The value of diagnostic tests is harder to measure than that of therapeutic interventions, however. Premarketing trials are typically aimed at determining accuracy, and insights about clinical utility often emerge only in the post-marketing setting. Uncertainty about new test interpretations may affect the adoption decisions of health care providers. For example, the AlloMap molecular expression test was developed to detect acute cellular rejection in heart-transplant patients. Premarketing studies suggested that while the AlloMap might have somewhat lower positive predictive value than biopsies, the fact that it is non-invasive gave it an advantage. After it was introduced, however, uncertainty about its clinical utility led many centers to use the test as an add-on and not a substitution—a common phenomenon with new diagnostic technologies. The reluctance to adopt the AlloMap as a substitute for biopsies was also influenced by the fact that cardiologists needed to become more comfortable interpreting the genomic information. In the area of pharmacogenetics, the integration of diagnostic tests and drugs poses special challenges because it will require that historically separate regulatory pathways be brought together. One successful example of such integration is HER-2 testing and Herceptin, where both products were approved through the fast-track process in the same week, with coordinated labeling. This case may have been relatively straightforward in that there was a clear relationship between the biomarker and drug response and the drug resulted in survival benefit for a life-threatening condition. With tests that have more ambiguity about the ultimate value of the information, rigorously conducted studies in the pre- and post-marketing stages will be even more important. Payers, who struggle with tradeoffs between costs and benefits, are exercising an increasingly important gatekeeper function through their coverage and reimbursement decisions. Although cost-effectiveness is not formally a coverage criterion for Medicare, many payers have adopted it as part of their decision-making process. Yet cost-effectiveness analyses of emerging novel technologies are challenging, partly because substantial innovation can be expected to take place after the technology goes to market. A strict adherence to a cost-effectiveness value such as $100,000 per life-year saved might eliminate some potentially valuable technologies before they have had the chance to prove their worth. In the case of genomic technologies, cost-effectiveness analyses need to incorporate post-marketing innovation and learning-by-using sensitivity analyses in a more systematic manner. At the same time, payer decision mak-
OCR for page 20
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary ing may need to become flexible enough to allow for short-term inefficiencies in order to understand and benefit from long-term value. Still, optimal learning takes time and experience, and payers may be understandably uncomfortable in underwriting such learning. This raises the important questions: What models can be used? And are there public–private partnerships that can be used to capture post-introduction learning more efficiently? Finally, the diffusion of genomic interventions is likely to be powerfully shaped by sociocultural factors. Even if genomic interventions are covered by insurers, patients may decide to pay out of pocket because of concerns about confidentiality and the potential for discrimination by employers and insurance companies. This, in turn, raises concerns about equity—for example, about lack of access to these technologies for those who do not have the means to pay. Another issue concerns the diffusion of tests that would identify predispositions to future disease for which there are no cures, only treatments with limited effectiveness and major side effects. Patients may vary greatly in their decisions about whether to have the genetic test for Huntington’s Disease, for example. Diffusion processes are fundamentally affected by patient preferences and by the public’s perception of the value of health-risk information. Gelijns concluded by saying that diffusion is a critical process by which the health, social, and economic rewards of an invention are ultimately reaped. Even more than that, however, diffusion is an integral part of the innovation process. It can be characterized as a learning process, and a fundamental aspect of learning is the reduction of uncertainty. Downstream learning can lead to changes in a technology or to refinements in its use. At the same time it poses new questions for basic and translational research and thereby enriches the ultimate payoff. The determinants of diffusion in genomics are probably very similar to those for other medical technologies. Diffusion depends not only on the benefits that a new intervention provides, but also, importantly, on the institutional environment in which a technology is imbedded. Patients, consumers, and physicians need to understand what to do with new probabilistic risk information; the FDA must decide how to deal with genetic diagnostic tests and how best to regulate diagnostic drug combinations; insurers need to gain comfort with the interpretation of cost-effectiveness analyses of emerging novel genomic technologies; and, finally, the larger policy world will need to deal with privacy and confidentiality issues and the potential for discrimination.
OCR for page 21
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary DISCUSSION Wylie Burke, M.D., Ph.D. Moderator A member of the audience commented that she believed Gelijns overestimated the use and the effect of cost-effectiveness analysis (CEA) in insurance decisions. The United Kingdom, she said, has adopted CEA and uses it for health-policy decisions, including those made by the National Health Service, but there is no mechanism in the United States for such analysis, and there is no systematic application of a quality threshold. While the Blue Cross and Blue Shield Association conducted a cost-effectiveness analysis regarding LVAD, it was done for educational purposes in order to understand the methodology. Califf commented that the lack of use of cost-effectiveness analysis illustrates the chaos that exists in decision making about innovations. Some innovations are blocked, while others go forward despite extraordinary costs, but it is very difficult to understand the basis upon which the decisions are made. Another audience member observed that it appears that as far as innovation is concerned, the health care system may reward small and relatively inconsequential changes and may sometimes create a disproportionate and negative response to rare events. Furthermore, the system prevents open exchange of information and creates many barriers to communication among affected parties—for example, between FDA and vendors or between those manufacturing or creating new devices or drugs and those who will be using them. This is the sort of situation described by the mathematician John Nash nearly 60 years ago—that is, that the optimal good is almost never achieved by the individual players optimizing their own individual results without being able to fully discuss how to jointly optimize the system (Nash, 1950). Could it be that the current health care system is so inhibited that the need to optimize individual results actually makes it impossible to introduce disruptive new technologies? Schulman responded by saying that markets evolve through a private process, and that private process is being choked. Porter wrote a book describing the different things that each of the actors in society can do to improve things in the health field—how hospitals could serve the needs of the public better, how physicians and insurers could do better, and so on (Porter, 2006). But there is nothing in the book, Schulman said, that explains why any of these actors would actually move from their current position. The critical issue, he continued, is the marketplace. What are the
OCR for page 22
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary levers? How can opportunities for leveraging and enforcing innovation be created in this very public market? Califf suggested that there are two important factors involved. First, there is an assumption that everyone agrees on what the ultimate state is or should be, which is not actually the case and which should be discussed further. If the goal were to optimize the longevity and quality of life of the American people as a population, that is not what we are now doing. But that may not be the goal everyone has in mind. Second, one must ask whether or not there actually is a need for disruption, as Schulman posits. If one makes a poor-quality transistor radio, the consequences are not great—the radio breaks, and it can easily be replaced. But if there is a poor-quality test that results in someone dying earlier than they otherwise might have, that should not be allowed to happen—a new test should not be allowed to enter the marketplace until studies have been conducted to show that it is worthwhile. Because the measurement of health status and health outcomes is so much more detailed and reliable than it once was, it is possible to measure what is being done and introduced. One audience member asked the speakers to go into greater detail about why they said the prospects for introducing genomic innovations into the marketplace are poor. Califf responded that part of the problem is that the regulatory pathway for introduction is unclear, and that lack of clarity discourages investors. Furthermore, genomic innovations target relatively small groups of people who can really benefit from the innovation. If the innovation concerns a disease such as cancer, investors will invest because the potential payoff is so large. For anything else, finding investors is a problem because the market is smaller. Gelijns said that one of the important issues is that the premarketing trials often focus on the accuracy of a test and that the ultimate clinical utility of these tests frequently emerges in the post-marketing setting. At that point one must deal with the issue of how best and most efficiently to obtain information about health and economic outcomes. Because individual stakeholders might not have enough incentive or means to conduct post-market testing, it is important to start thinking about new models of cooperation, such as public–private partnerships, that will pull together the various parts of the system to generate needed information and to improve the innovation process. One questioner said that because he comes from a public health background he would like to see discussion about how the translation and diffusion of innovations take into account the end of the pathway—that is, improving the public’s health. In many areas of medicine and public health it is known that if a particular action is taken, thousands of deaths in the population can be prevented, yet those things are not being done. It takes years to implement and diffuse proven innovations into practice.
OCR for page 23
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary On the other hand, there are new technologies, such as the genome-based technologies, that have uncertain added value compared with what is currently being done. These technologies are intended to replace or be added to the things that we know should be done but that are not actually being translated into practice. So the question is, is there a process or an organizing principle that helps sort out or distinguish what is ready to be introduced from what is not? Furthermore, given the complicated schema of translation and the multiple factors and players that are involved in this process, where is the role of the evidence-based guideline? Schulman responded that, intellectually, the use of clinical-practice guidelines for genomic innovations is an exciting area. There is still relatively little clinical information available, however, so how these guidelines will fit into the marketplace is uncertain. Califf agreed with the questioner that there are things we know now that could be used for better health or treatment. For example, a person admitted to a hospital with an acute coronary event has a 33 percent likelihood of getting the wrong dose of any thrombotic drugs that are prescribed. A simple serum creatinine in body mass index will give information about the correct dose, but it is often not used. Yet there is discussion of diffusing even more sophisticated genomic technology into outpatient settings or unsupervised settings. These new tests should not be unleashed upon the public without evidence that they will help rather than harm. On the other hand, when there is an important disruptive technology that can make a big difference, there should be some special approach that allows people to develop the evidence with some protection while a determination is made about whether the innovation is valuable. Gelijns said that a major question is how to create incentives for gathering information as the technology keeps changing in the post-marketing setting. Another questioner, stuck by the idea of post-market innovation and how much is learned when something is put into practice or use, asked if expanding regulations and the increasing drive for evidence-based practice of medicine are going to squelch innovation. Schulman said that one argument is that we need at least some level of evidence on the technology innovation side that there is some effectiveness. Then there is also the need for service innovation. Today’s system is costly and less effective than it should be, therefore service innovation is necessary. Califf stated that creativity is needed both in customizing (some say personalizing) medical care for patients and in delivering services. People running health systems find it impossible, given today’s finances, to actually do what people need to have done in order to make them healthier, he
OCR for page 24
Diffusion and Use of Genomic Innovations in Health and Medicine: Workshop Summary said. Furthermore, companies, in order to make money, study the wrong things because of the way incentives are structured. If one could create a more streamlined system and actually define the value of the products as opposed to developing marketing schemes that are tangential to the value, one could focus on defining what actually works. Gelijns mentioned that the history of pharmaceutical innovation is full of new and often unexpected indications of use that emerge in the post-marketing setting. What is needed may be a more streamlined process for looking at evidence, not only of unexpected side effects, but also of unexpected benefits. One important issue is how much room to leave for experimentation. One audience member said that he thought that a major player in innovation and translation—the pharmaceutical companies—was being ignored. In 1998 a drug that had a 4 percent complication rate of hypersensitivity syndrome (abacavir) was released with an accelerated approval that required risk-management studies to be conducted. As a result, some early-stage pharmacogenetics were conducted. At the time, it was the only product in its class. Other products in the class have now entered the marketplace, and each of them has a different type of adverse event. From the company’s point of view, if there was a highly accurate test that demonstrated that the adverse event for its drug could be avoided, the company would have a competitive advantage to getting that put on the label. That is, in fact, what happened. There is now a test that can identify, with greater than 99 percent specificity, the people who will suffer hypersensitivity syndrome if they take abacavir. For the first time there is a diagnostic test for a drug allergy. Month by month the sales of this test, which did not come out simultaneously with the drug, have quadrupled, according to the audience member. Yet this information has not been published in any of the scientific journals. This example illustrates the fact that if there is a competitive situation and market share is at stake, a huge incentive exists to make the investment needed to implement the test, even if there is no reimbursement for the test. It might even be considered unethical to give the drug without testing. It is important, the audience member concluded, not to restrict thinking about incentives to academia and the government, but rather to expand incentives to include those who can change the system, such as pharmaceutical companies.